From Reactive Networks to Cognitive Cities: A SMILE-Based Digital Twin Roadmap for 6G, Adaptive Urban Intelligence, and Holographic Societies (2025–2040)

Author: Nicolas Waern Affiliation: WINNIIO AB / Life Atlas ORCID: 0009-0001-4011-8201 Date: April 2026 License: Creative Commons Attribution 4.0 International (CC BY 4.0)

Abstract

The convergence of sixth-generation wireless networks, multi-domain digital twins, edge-native intelligence, and extended reality platforms marks a civilizational inflection in how cities sense, reason, and act. This paper presents a structured 15-year roadmap (2025–2040) organized through the S.M.I.L.E. (Sustainable Methodology for Impact Lifecycle Enablement) framework, tracing the evolution from today's reactive, siloed smart-city deployments through adaptive, data-integrated urban systems toward fully cognitive cities — twin-native, anticipatory, and capable of self-governance at the neighborhood scale. We argue that 6G does not merely accelerate 5G; it constitutes the nervous system of a fundamentally different kind of city: one in which every physical system has a digital counterpart, every citizen has a sovereign personal guardian, intelligence is ambient rather than centralized, and presence is no longer tied to location.

The SMILE methodology structures this evolution across six concentric phases: Reality Emulation, Concurrent Engineering, Collective Intelligence, Contextual Intelligence, Continuous Intelligence, and Perpetual Wisdom. These phases are not sequential waterfall stages but iterative loops that different deployments traverse at different velocities. Grounding the framework in O-RAN Alliance specifications, the Life Programmable Interface (LPI) as sovereign consultation layer, 3D scene reconstruction via Gaussian splatting and neural radiance fields, vehicle-to-everything (V2X) communication standards, multi-agent systems architecture, extended reality overlays, and post-quantum security architecture, we define concrete capability milestones, governance requirements, and interoperability conditions for each phase. We introduce three new architectural pillars relative to the prior version: edge-native intelligence as a hard sub-millisecond constraint (not a cloud optimization), the Internet of People as the human-sensor layer that makes biological data a first-class city input (with full consent architecture), and holoportation as the social infrastructure of the cognitive city. The paper concludes with a discussion of the digital divide, ethical risks of ambient intelligence, and an economic model for community-owned cognitive infrastructure that aligns incentives across citizens, operators, and municipalities.

Keywords: 6G, cognitive cities, digital twin, SMILE methodology, Life Programmable Interface, adaptive urban intelligence, holographic communication, holoportation, XR, edge-native intelligence, V2X, vehicle-to-everything, Internet of People, autonomous agents, multi-agent systems, MAGS, energy harvesting, biodigital twin, data sovereignty, O-RAN, Gaussian splatting, post-quantum cryptography, personal guardians, differential privacy, epigenetics, energy resilience, 3D bioprinting

1. Introduction — The Convergence Imperative

1.1 From Reactive Networks to Anticipatory Intelligence

In 2025, the world's most advanced wireless networks remain fundamentally reactive. A cell degrades, a KPI threshold breaches, an engineer is paged. The optimization event occurs after the user has already experienced failure. The intelligence promised by open, programmable radio access networks — the RAN Intelligent Controller, xApps, rApps, intent-based automation — is constrained by the absence of a continuously updated, physics-aware model of the physical world the network serves.

At the same time, cities instrumented by a decade of smart-city investment — sensors embedded in road surfaces, air quality monitors on lampposts, cameras on every intersection — generate data volumes that municipal systems cannot absorb, interpret, or act upon in time to be useful. The data is there. The models to make sense of it are not. The connections between the network layer, the urban layer, and the human layer have not been made.

6G changes this not merely through higher bandwidth and lower latency but through a conceptual reframing: from connectivity as a service to intelligence as infrastructure. The ITU-R IMT-2030 framework [1] positions 6G as an integrated sensing-and-communication system, embedding the network in the physical world as a distributed sensor fabric. That fabric, combined with digital twin technology capable of ingesting its outputs and running forward-looking simulations, produces something qualitatively new: a city that knows what is happening before it happens.

This paper argues that realizing this potential requires architectural discipline that no single technology layer can provide on its own. The organizing framework is SMILE — Sustainable Methodology for Impact Lifecycle Enablement — a six-phase, three-perspective methodology that sequences how intelligent urban systems are designed, deployed, and evolved [2]. SMILE is not a vendor framework and not a standards body specification. It is a methodology for outcome-first engineering across multi-stakeholder ecosystems, compatible with and applicable to O-RAN Alliance architectures [3][4], the Open & Agile Smart Cities (OASC) Minimal Interoperability Mechanisms (MIMs) [5], and ISO 37120 smart city indicators [6].

1.2 The SMILE Methodology as Organizing Principle

S.M.I.L.E. stands for Sustainable Methodology for Impact Lifecycle Enablement. It defines six concentric phases through which any complex, multi-stakeholder digital twin evolves:

1. Reality Emulation — Establish a shared reality canvas. Map stakeholders and their relationships using Actor-Network Theory [7][8]. Build the virtual representation before instrumenting the physical. 2. Concurrent Engineering — Define scope collaboratively. Build the Minimal Viable Twin (MVT). Validate hypotheses before committing to full-scale deployment. 3. Collective Intelligence — Connect physical sensors. Align KPIs across stakeholder domains. Create the ontological foundation — the shared vocabulary — that makes cross-domain data interoperable. 4. Contextual Intelligence — Connect everything to the twin. Establish command-and-control interfaces. Enable real-time decision support at the point of action. 5. Continuous Intelligence — Deploy prescriptive maintenance and AI-driven prognostics. Simulate everything. Create the closed-loop automation that makes the system self-optimizing. 6. Perpetual Wisdom — Enable ecosystem-wide knowledge sharing. Implement circular strategies. Make the methodology itself a replicable asset for other cities, networks, and communities.

Three perspectives cut across all phases:

• From People: How does this serve individual humans — their sovereignty, their legibility, their ability to override?
• From Systems: How do technical components interact, interoperate, and degrade gracefully?
• From Planet: What are the environmental, equity, and long-term sustainability implications?

SMILE is the structural backbone of this paper. Each section maps to one or more SMILE phases. The roadmap is not a prediction of what will happen; it is a description of what must be built, in what order, to move from reactive networks to cognitive cities without losing sovereignty, equity, or explainability along the way.

1.3 Paper Contributions and Structure

This paper makes six primary contributions:

1. A reframed evolution model — Smart City → Adaptive City → Cognitive City — mapped to 6G capability milestones and grounded in verifiable technical developments. 2. A SMILE-structured implementation roadmap with concrete phase gates, data governance requirements, and interoperability conditions. 3. A description of the Life Programmable Interface (LPI) as the sovereign consultation layer that must sit between AI optimization systems and physical infrastructure. 4. An architectural treatment of edge-native intelligence as a hard latency requirement — not an optimization — including V2X and multi-agent coordination. 5. An introduction of the Internet of People paradigm: the human as the most important sensor and actuator in the cognitive city, with full consent and sovereignty architecture. 6. An economic model for community-owned cognitive infrastructure aligned with the political economy of 6G deployment.

Section 2 grounds the roadmap in the three-era city evolution model and positions 6G and LPI as the two enabling infrastructures. Sections 3 through 5 map SMILE phases 1–6 to the 2025–2040 timeline. Section 6 describes the LPI architectural framework in detail. Section 7 discusses challenges, ethics, and economic models. Section 8 concludes with future work.

2. Background: From Smart to Cognitive Cities

2.1 Era 1 — Smart Cities: Sensor-Driven, Siloed (2010–2020)

The smart city era of 2010–2020 was defined by instrumentation: sensors embedded in infrastructure, data transmitted to central platforms, dashboards produced for municipal analysts. The paradigm was largely extractive — data flowed from the physical world to command centers — and largely siloed, with transport data in one system, energy in another, health in a third.

Batty (2013) characterized the first-generation smart city as a system of systems lacking integration — rich in data, poor in understanding [9]. Mostashari et al. (2011) noted that the dominant smart-city architecture of the period — centralized data aggregation with top-down control — was antithetical to the complex adaptive nature of urban systems [10]. Cities are not machines to be optimized; they are ecosystems to be understood.

The evidence bore this out. Despite billions of dollars in smart-city investment globally, the majority of first-generation deployments failed to demonstrate measurable improvement in citizen outcomes. Data was collected that was never acted upon. Dashboards were built that were never used. Interoperability between vendor platforms was absent, making cross-domain insight structurally impossible.

2.2 Era 2 — Adaptive Cities: Data-Integrated, Responsive (2020–2030)

The 2020–2030 period marks the transition to adaptive cities: urban systems in which data from multiple domains is integrated at the platform level, enabling responsive (if not yet predictive) management. Enabling developments include: the maturation of digital twin platforms for urban environments (Singapore's Virtual Singapore [11], Helsinki's urban twin [12], Amsterdam's 3D city model), the deployment of 5G networks with programmable interfaces, and the emergence of open standards for city-scale data sharing including the FIWARE NGSI-LD context broker and the OASC MIMs framework [5].

The adaptive city can respond to events in near-real time: rerouting traffic when an incident occurs, adjusting public transport frequency based on crowd density, dimming street lighting in low-activity areas. But response remains the operative word. The system reacts to what has happened, not to what is about to happen. The data integration layer has been built; the forward-looking intelligence layer has not.

5G provides the connectivity substrate for the adaptive city, and O-RAN disaggregation — formalized through deployments including Rakuten Mobile's cloud-native network in Japan [13][14] — creates the programmatic surface through which network intelligence can be composed from third-party applications. But 5G, as deployed in the adaptive-city era, is a communication technology. It does not yet sense the physical world; it does not yet reason about it.

2.3 Era 3 — Cognitive Cities: Twin-Native, Anticipatory, Self-Governing (2030–2040)

The cognitive city is the destination of this roadmap. Mohammadi and Al-Fuqaha (2018) define cognitive cities as urban environments that employ AI and machine learning to continuously learn from data, adapt to changing conditions, and make proactive decisions that improve citizen welfare [15]. We extend this definition in four directions:

• Twin-native: Every significant physical system in the cognitive city has a continuously synchronized digital counterpart. The twin is not a visualization tool; it is the authoritative operational model of the city.
• Anticipatory: The city acts on predictions, not observations. Resource allocation, traffic management, network capacity, and emergency response are governed by what the twin expects to happen in the next 15 minutes to 72 hours, not by what happened in the last 15 minutes.
• Self-governing at the neighborhood scale: Governance authority is distributed to the community level. Neighborhoods can inspect, audit, and override AI decisions affecting their residents. Sovereignty is not a policy aspiration; it is an architectural property.
• Presence-decoupled: Physical location no longer constrains participation. Through holoportation and XR interfaces, people inhabit the cognitive city across geographic distances, and the city's AI systems account for both physical and virtual presence in their resource allocation and social coordination functions.

2.4 6G as the Nervous System of Cognitive Cities

The ITU-R IMT-2030 framework defines 6G around six usage scenarios: immersive communication, hyper-reliable and low-latency communication, massive communication, ubiquitous connectivity, integrated sensing and communication (ISAC), and integrated AI and communication [1]. The last two are architecturally new relative to 5G.

ISAC means that the 6G radio is simultaneously a communication medium and a sensor: it can reconstruct the physical environment from reflected radio waves, detect human presence and motion, and infer the state of objects without dedicated sensor hardware. This transforms the 6G network from a pipe into a perceptual organ — one that feeds continuous environmental state data into the city's digital twin.

Integrated AI and communication means that AI inference runs in the network itself, not merely in devices or cloud servers that communicate over the network. Intelligence is distributed to the radio unit, the distributed unit, the edge server. The city's cognitive capacity is proportional to the density of its 6G deployment.

Nokia Bell Labs' 6G vision [16], Samsung's 6G white paper [17], and the Hexa-X project's architectural contributions [18][55] all converge on a similar picture: 6G as a programmable, sensing-capable, AI-native infrastructure that is inseparable from the digital twin layer it feeds.

One implication demands emphasis. Sub-millisecond decisions — autonomous vehicle collision avoidance, holographic motion-to-photon latency, surgical robotic control — cannot tolerate the 30–80ms round-trip latency of a cloud data center 500km away. They require inference at the edge: on the vehicle, at the roadside unit, in the building's own compute fabric. This is not a cost optimization. It is a physical constraint. The speed of light in fiber is approximately 200,000 km/s; the round trip to a centralized cloud takes time that the application cannot afford. Edge-native intelligence is thus the architectural floor of the cognitive city, not a premium feature.

2.5 The Life Programmable Interface as Sovereign Consultation Layer

The cognitive city creates a governance challenge that no technical standard alone can solve: when an AI system operating a city's traffic management, network optimization, and emergency response simultaneously can take actions affecting millions of people, who is responsible for those actions? How are they explained, audited, and overridden?

The Life Programmable Interface (LPI) is the answer developed within the Life Atlas architecture [2]. The LPI is a sovereign consultation layer that sits between AI optimization systems and the physical infrastructure they control. Every automated action — before it is applied to a physical system — must pass through the LPI, which enforces three invariants: explainability (human-readable rationale), overridability (time-bounded human veto), and sovereignty (data subject rights preserved at every step).

LPI draws on NeMo Guardrails for conversational AI safety [63], OpenLLMetry/OpenTelemetry for full observability of every LLM call, and differential privacy [19] for aggregate data releases. At the personal scale, LPI manifests as the individual's guardian — the Life Atlas personal twin that monitors, reasons, and acts on behalf of a specific person, never on behalf of the city's aggregate optimization function at the expense of that person's interests.

This is not merely an ethical aspiration. It is an architectural requirement for the cognitive city to achieve democratic legitimacy. Cities that deploy ambient intelligence without sovereign consultation layers will face — and deserve — regulatory intervention. The LPI is how that intervention is pre-empted by design.

3. SMILE Phases 1–2: Reality Emulation and Concurrent Engineering (2025–2028)

3.1 Phase 1 — Reality Emulation: The Shared Canvas

The first SMILE phase is not a technical deployment. It is a sense-making exercise. Before any sensor is connected, any algorithm trained, or any interface built, stakeholders must agree on what reality they are trying to model and why.

Reality Emulation draws on Actor-Network Theory (ANT) as developed by Latour (2005) and Callon (1986) [7][8]. ANT rejects the distinction between social and technical actors: a traffic signal, a municipal planning regulation, a neighborhood association, a 6G antenna, and a city councilor are all actors in the same network. The first step of SMILE is mapping this network: who are the actors, what are their interests, what mediates their relationships, and where are the controversies that will determine whether the twin succeeds or fails?

In the telecom context, Reality Emulation produces:

• A stakeholder map that includes operators, regulators, equipment vendors, municipalities, citizens, and civil society organizations — with their data-sharing constraints and sovereignty requirements made explicit before architecture is designed.
• A virtual first principle: the initial twin is built from available data (GIS, existing sensor feeds, historical network telemetry, planning documents) before any new infrastructure is deployed. The Minimal Viable Twin is functional before a single new sensor is installed.
• A shared ontology draft: the common vocabulary that will govern what concepts — "cell," "neighborhood," "citizen," "service level," "energy budget" — mean across all participating organizations.

Applied to Open RAN environments, Phase 1 produces a network twin populated with existing NMS/OSS data, GIS-derived environment models, and historical MDT (Minimization of Drive Test) reports. This virtual twin is not high-fidelity; it is high-clarity. It makes visible the relationships that will govern the more complex system to come.

3.2 Phase 2 — Concurrent Engineering: The Minimal Viable Twin

The Minimal Viable Twin (MVT) is the SMILE equivalent of the MVP in product development, with one critical difference: the MVT must be grounded in real physics, not mocked data. A mock MVT is not a twin; it is a wireframe with delusions of grandeur.

Concurrent engineering — the second SMILE phase — means that the outcome specification, the technical architecture, and the governance framework are developed simultaneously, in the same room, with all stakeholders present. The sequential approach — first decide what to build, then decide how, then decide who governs — produces systems that fail at the governance layer because governance was an afterthought.

For the 2025–2028 period, the MVT architecture centers on:

Outcome specification: Before any algorithm is trained, the operator specifies the quality-of-experience outcome the twin must achieve — expressed as a measurable, stakeholder-agreed KPI. For a telecom operator: handover success rate above 99.5% at the 95th percentile across all cells. For a municipality: pedestrian-scale air quality index below WHO threshold in residential neighborhoods by 2027.

Virtual Twin Lab for RAN Optimization: Drawing on the Rakuten Mobile case study [13][14], the MVT for telecom is a simulation-first environment in which xApp policies for the Near-RT RIC are trained and validated before production deployment. The O-RAN Alliance AI/ML workflow specifications (O-RAN.WG2.AIML) [4] define the deployment pipeline from twin training to production RIC. The twin provides the gymnasium-compatible reinforcement learning environment; the xApp is the policy; the E2 interface carries the control actions.

Security architecture — zero-trust, edge-native, PQC-ready: Security is not a Phase 3 concern. It is designed in Phase 2. Zero-trust network access for all twin components, post-quantum cryptography (NIST PQC standards [20]) on all data-at-rest and long-term key exchanges, and edge-native architecture that processes sensitive data locally — eliminating centralized attack surfaces — are non-negotiable Phase 2 requirements. The EU AI Act [21] and GDPR [22] compliance requirements are mapped to specific architectural controls during Concurrent Engineering, not retrofitted after deployment.

Self-healing network design: Phase 2 architecture bakes resilience in as a design primitive, not a recovery feature. Self-healing network topology — the ability to detect node or link failure and automatically reroute traffic before users notice the disruption — is specified as a Phase 2 outcome requirement. Current research demonstrates that software-defined networking combined with real-time digital twin state awareness can reduce mean-time-to-recovery for network faults from minutes to sub-second timescales [64]. The twin continuously monitors network health, predicts degradation trajectories, and pre-positions backup paths before failure occurs.

Failure mode inventory: The MVT architecture includes explicit simulation of failure modes: backhaul disruption, sensor dropout, adversarial inputs, model drift. A system that works only in optimal conditions is not a twin; it is a demo.

3.3 Outcome-First vs. Data-First Failure Modes

The most common failure in digital twin projects — confirmed across numerous case studies in the research literature [23] and in operational practice — is the data-first trap: organizations build enormous data pipelines, aggregate petabytes of sensor data, and then discover that none of it is organized around questions anyone actually needs answered. The twin becomes a data lake with a 3D skin.

SMILE's Phase 2 mandates the opposite order: define the question first, then determine what data is required to answer it, then and only then deploy the infrastructure to collect that data. This is not merely methodological preference; it is how you maintain budget discipline, stakeholder alignment, and production velocity when crossing organizational boundaries in a multi-stakeholder urban system.

4. SMILE Phases 3–4: Collective and Contextual Intelligence (2028–2033)

4.1 Phase 3 — Collective Intelligence: Connecting the Physical World

The transition from virtual twin to digital twin — from planning-time simulation to real-time synchronization — is the Phase 3 step change. Collective Intelligence connects the physical world to the twin: sensors report state, KPIs are measured continuously, and the shared ontology created in Phase 1 is instantiated in a knowledge graph that makes cross-domain queries possible.

Real-time dynamic twins with sub-second state updates are the defining Phase 3 capability. The O-RAN open interface stack provides the data pathways: E2 (per-UE/per-cell measurements at 10ms minimum), O1 (management plane configuration and fault state), and A1 (policy from Non-RT RIC to Near-RT RIC) [3][4]. The twin state engine consumes these streams and propagates updates through a dependency graph — recomputing propagation models, interference estimates, and traffic predictions only for the components whose inputs have changed.

KPI alignment across stakeholder domains is a governance achievement, not a technical one. When the network twin's "coverage quality" KPI must be reconciled with the city twin's "pedestrian digital inclusion index" and the health twin's "emergency response latency guarantee," the reconciliation happens in the shared ontology — not in ad-hoc data transformations. The Digital Twin Consortium's reference architecture [24] and the ISO 23247 digital twin manufacturing framework [25] provide vocabulary. SMILE's ontology creation process provides the governance ritual that gets stakeholders to agree on definitions before they become contested.

Ontology Factories to AI Factories: The shared ontology created in SMILE Phase 3 is not just a data schema. It is the foundation of an AI Factory — the infrastructure through which domain knowledge is systematically encoded in models that can reason across domains. Latour's distinction between intermediaries (entities that transport meaning without transformation) and mediators (entities that transform meaning in transit) [7] maps directly to the distinction between data pipelines (intermediaries) and ontology-grounded AI models (mediators). The Collective Intelligence phase builds mediators, not just pipelines.

4.2 V2X: Cars That See Through Corners

Among the most consequential Phase 3 capabilities is vehicle-to-everything (V2X) communication: the integration of autonomous and connected vehicles into the city twin as both data producers and acting agents within it. Standardized through 3GPP Cellular V2X (C-V2X, specified in 3GPP Release 16 and Release 17) [65] and ETSI ITS (Intelligent Transport Systems) [66], V2X enables vehicles to exchange trajectory, speed, and intent data with other vehicles, roadside infrastructure, pedestrians' devices, and the city digital twin simultaneously.

This dissolves the line-of-sight constraint that defines the safety horizon of traditional vehicle sensors. A car approaching an intersection cannot see through the building on the corner. But the city twin can. It knows the trajectory of the cyclist coming the other way, the pedestrian stepping off the curb, the ambulance approaching from three blocks away. Through V2X, it tells every vehicle in the vicinity — not as a warning to react to, but as a prediction to plan around. Predictive collision prevention replaces reactive obstacle detection. The twin's situational awareness becomes the vehicle's situational awareness.

The safety implications are substantial. Current research demonstrates that C-V2X cooperative awareness reduces intersection accident rates significantly in simulation environments [65]. At the Phase 3 horizon (2028–2030), early V2X deployments in urban corridors — drawing on both DSRC (Dedicated Short-Range Communication) infrastructure already deployed in several cities and C-V2X over 5G/6G networks — will produce the longitudinal datasets needed to validate these safety models in real urban conditions.

The city twin does not merely passively receive V2X telemetry. It actively orchestrates: pre-positioning traffic signal phases, alerting pedestrian crossings, advising transit operators, and — critically — pre-positioning network capacity. When 40 autonomous vehicles enter a block simultaneously, the network twin knows before they arrive. Beam resources are pre-allocated. The network does not react to demand; it anticipates it.

4.3 3D Scene Reconstruction for RF and Urban Intelligence

Accurate modeling of the physical environment — the geometry and material composition of buildings, the density of vegetation, the positions of vehicles and pedestrians — is the enabling substrate for both RF propagation modeling and urban digital twins. Traditional approaches require expensive LiDAR surveys and manually maintained GIS databases.

Neural Radiance Fields (NeRF), introduced by Mildenhall et al. (2020) [26], demonstrated that photorealistic, geometrically accurate 3D scene representations can be reconstructed from collections of 2D images. 3D Gaussian Splatting (3DGS), introduced by Kerbl et al. (2023) [27], achieves comparable quality at real-time rendering speeds — two to three orders of magnitude faster than NeRF — through an explicit Gaussian primitive representation. This performance characteristic is critical for digital twin applications where the scene model must be updated as the physical environment changes.

For urban digital twins, the practical implication is that street-level video from operator maintenance vehicles, municipal infrastructure cameras, and citizen devices (with appropriate consent frameworks) can continuously update the city's 3D model without specialized survey equipment. For network twins, this means that the propagation environment feeding the ray-tracing simulator is updated weekly or monthly from commodity data, replacing drive test campaigns that cost metropolitan operators days of field time per cycle [28].

Integrated Sensing and Communication (ISAC) in 6G extends this further: the radio itself becomes a scene reconstruction sensor, using reflected millimeter-wave and sub-THz signals to build and maintain the 3D environment model as a byproduct of communication. The network senses its physical context continuously, without dedicated sensing hardware [1][16].

4.4 XR Overlays: The City Becomes Legible

Phase 3 also introduces the first consumer-facing interface to the cognitive city's intelligence: extended reality overlays. By 2028, XR glasses — in their first-generation consumer form, following the trajectory established by platforms such as Microsoft Mesh [67] and Meta's Codec Avatars research [68] — will allow citizens to see the city's data layer superimposed on the physical world. Air quality gradients visible as color overlays. Energy flow in a building's facade. Network signal quality mapped onto the street. The health index of the neighborhood park, updated in real time.

This is not novelty. It is legibility. The cognitive city collects enormous amounts of data about its own state; XR overlays are the interface through which citizens can read that state without specialized training. A person walking through a neighborhood with poor air quality sees it. A cyclist choosing a route sees the network coverage and the predicted traffic load. The city's intelligence stops being opaque.

Gamification and incentivized campaigns run natively in this interface layer. "Walk this route and earn LIFE tokens — it reduces your carbon exposure by 30% and the city's peak congestion by 0.1%." "Turn off your building's climate system for the next two hours during grid stress and watch your energy twin's contribution to the city's resilience in real time." Every action becomes a negotiation between personal interest and collective benefit — visible, explainable, and rewarded. The differential privacy guarantee [19] ensures that personalization never requires identification: the system knows enough to tailor the incentive, never enough to build a surveillance profile.

4.5 Cross-Domain Twin Convergence

The most strategically significant SMILE Phase 3 development is not within any single domain; it is the convergence of twins across domains. A network twin that knows about pedestrian mobility predicts demand differently from one that does not. A city twin that incorporates network QoS data plans public space differently. A health twin that can reason about network availability guarantees for emergency services operates at a different level of reliability than one that treats connectivity as a given.

Cross-domain convergence in the 2028–2033 period includes:

• Network + Urban: Municipal digital twins (Singapore [11], Helsinki [12], Zurich [29]) integrated with operator network twins, enabling mobility-aware RAN optimization and network-aware urban planning.
• Network + Mobility: Autonomous vehicle platforms and smart mobility systems sharing trajectory predictions with network twins, enabling proactive handover preparation and capacity pre-positioning. V2X communication (3GPP C-V2X [65], ETSI ITS [66]) provides the protocol substrate.
• Network + Health: The Life Atlas biodigital twin architecture [2][53] integrated with network QoS guarantees, enabling health monitoring applications that can assert connectivity requirements as first-class SLA parameters.
• Network + Energy: The energy twin and the telecom twin share real-time state. When renewable production drops below forecast — a solar irradiance event, a wind lull — the city twin pre-adjusts demand: deferring non-critical computation, pre-cooling buildings before they need it, signaling V2X-connected vehicles to charge at off-peak times. The cognitive city's energy and communication systems are not separate domains; they are coupled layers of the same physical reality.

Cross-domain convergence requires federated data governance. Query federation over data centralization is the architectural principle: each domain retains sovereignty over its data; the inter-twin query layer retrieves derived insights on demand, applying the data owner's access policies at query time. ETSI MEC specifications [30] and emerging OASC MIM implementations [5] provide the protocol substrate.

4.6 Edge Intelligence: Local Inference, Sovereign Data Processing

The SMILE Phase 4 transition — from connected systems to contextual intelligence — requires that intelligence operates at the point of action, not only in the cloud. Edge computing, as defined by Shi et al. (2016) and Satyanarayanan (2017) [31][32], brings computation to the data rather than data to the computation. For urban systems where latency, privacy, and resilience are all constraints, edge inference is not an optimization; it is a requirement.

Consider the numbers concretely. The round-trip latency from an autonomous vehicle to a centralized cloud server is 30–80ms on a good day. A vehicle travelling at 100 km/h covers 2.8 meters in that time. The 3GPP specification for Vehicle Platooning and Advanced Driving (Release 16) requires end-to-end latency of 3–10ms [65]. Only edge inference — running on the vehicle itself, on a roadside 6G node, or at a Multi-access Edge Compute (MEC) facility at the base of the nearest cell tower — can satisfy this requirement. This is why NVIDIA's Jetson ecosystem [69] and Qualcomm's AI Engine platform [70] are not edge curiosities; they are the computational substrate that makes autonomy safe.

The SMILE principle of edge-native first means: design the system so that it functions at full capability on the local node, and treats cloud resources as additive bandwidth and compute, not foundational infrastructure. Ollama and Bonsai-class local inference environments — edge-viable LLM infrastructure deployable on commodity hardware — enable the twin's AI reasoning to operate on a sovereign community node without cloud connectivity for every inference.

Self-learning systems amplify this compounding over time. The city twin does not merely operate its models; it retrains them continuously from the gap between prediction and observation. Every interaction — a pedestrian detouring an unexpected obstruction, a building system responding to an unusual weather pattern, a network cell recovering from interference — is a training signal. Every day, every season, the system's models improve. The cognitive city of 2035 is not the cognitive city of 2028; it is the cognitive city of 2028 after seven years of continuous self-improvement on real operational data.

4.7 Personal Guardians: Life Atlas as the Individual's Twin Interface

The cognitive city's ambient intelligence creates a risk that is not primarily technical: the individual becomes a subject of the system's optimization function rather than its beneficiary. Traffic flows smoothly; a specific pedestrian's medical emergency is deprioritized because she is in a low-density area that does not affect aggregate KPIs.

Life Atlas's personal guardian architecture is the individual-scale counterpart to the city-scale cognitive infrastructure. Each person's guardian is a continuously updated model of that person's state — biological, environmental, social, spatial — that operates exclusively in their interest, mediated by the LPI. The guardian does not optimize for the city; it advocates within the city's systems for the individual's sovereign interests.

The guardian architecture connects to the network twin through the LPI's quality-of-service assertion interface: the guardian can declare "this user requires sub-100ms network latency for the next 4 hours" (because she is using a real-time health monitoring device), and the network twin incorporates this as a hard constraint in its resource allocation. The network serves the person; the guardian ensures the network knows what the person needs.

This is the personal-scale instantiation of SMILE's "From People" perspective: the question is not only "how does this system serve society in aggregate?" but "how does this system serve this specific person, and who answers for it when it fails?"

The individual's guardian also negotiates on their behalf across domains simultaneously. Bandwidth allocation with the network twin. Route recommendation with the mobility twin. Air quality routing with the urban environment twin. Energy credits with the energy twin. The person specifies goals — I want to arrive at this meeting rested, with minimal pollution exposure, on time — and the guardian negotiates the city's resources to realize those goals. The city is a platform for sovereign human intention, not a system that humans navigate alone.

5. SMILE Phases 5–6: Continuous Intelligence and Perpetual Wisdom (2033–2040)

5.1 Phase 5 — Continuous Intelligence: Simulate Everything

The Continuous Intelligence phase is where the twin becomes the authoritative operational model of the city. It no longer merely reflects what is happening; it prescribes what should happen, simulates the consequences before acting, and learns continuously from the gap between its predictions and observed outcomes.

Self-aware networks and predictive demand modeling: By 2033, the network twin has accumulated years of correlated observations — network state, user mobility, application demand, environmental conditions, event calendars. This longitudinal dataset enables demand anticipation at 5-minute resolution across the coverage area, including event-driven pre-positioning (the twin knows that 60,000 people will arrive at a stadium in 90 minutes and begins beam reconfiguration one hour before) and application-aware QoS pre-provisioning based on learned demand signatures.

Autonomous agents negotiating the city: Phase 5 is where agent-based modeling reaches full operational scale. Macal and North (2010) demonstrated that agent-based models — systems of interacting autonomous decision-makers — can capture emergent collective behavior in complex urban systems that equation-based models cannot represent [71]. Bonabeau (2002) formalized the methodological foundations of multi-agent simulation [72]. By Phase 5, the cognitive city runs millions of agents in continuous parallel simulation: every vehicle, every building system, every pedestrian flow cluster, every network element — each with its own state, its own decision rules, and its own Perceive/Plan/Act/Reflect cycle.

Autonomous vehicles in Phase 5 do not merely communicate with each other through V2X; they negotiate. A vehicle approaching an intersection simultaneously considers: the trajectory of twelve other vehicles within V2X range, the pedestrian crossing signal, the emergency vehicle two blocks away requesting priority, and its own occupant's calendar — which suggests that arriving two minutes late at a lower speed costs nothing, while maintaining speed requires a sequence of rapid maneuvers that stress the city's energy grid at peak load. The vehicle's agent negotiates a cooperative outcome with eleven other vehicle agents and the intersection's city-twin agent — in milliseconds, without human intervention, in a way that optimizes for all of them simultaneously.

This is not science fiction. It is the logical extension of cooperative driving research already underway under 3GPP Release 17 and ETSI ITS G5 [66], extended to the 6G infrastructure and agent-architecture capabilities described in this paper.

Multi-agent systems architecture — MAGS-inspired: The city's AI architecture follows a Perceive/Plan/Act/Reflect (PPAR) cycle at every scale: vehicle agents, building agents, neighborhood agents, city-wide policy agents. Each agent maintains its own world model, generates candidate actions, selects actions based on current policy and governance constraints, and reflects on outcomes to update its models. The LPI operates as the governance layer above all agent action: no agent can take an action that violates sovereignty constraints, even if that action would improve the aggregate optimization objective. People in control — always, structurally, not just aspirationally.

Holographic society infrastructure requirements: ITU-T Study Group 16 has begun standardizing holographic communication [33], with requirements that exceed 5G capabilities: 1–10 Gbps per session for uncompressed volumetric video, end-to-end motion-to-photon latency below 10ms, packet delay variation below 1ms, and five-nines packet delivery reliability [34]. These requirements demand the combination of 6G air interface advances, dense small cell deployment, and intelligent traffic engineering — all of which presuppose a continuously updated digital twin as the traffic engineering substrate. No reactive system can maintain sub-millisecond jitter guarantees across a dynamic urban environment.

Holoportation: presence without location: By the Phase 5 horizon, holographic communication infrastructure [33][67] reaches sufficient maturity that holoportation — real-time, photorealistic volumetric representation of a remote person in a local physical space — becomes a functional social infrastructure. The technology builds on Microsoft's Mesh platform [67] and Meta's Codec Avatars research program [68], which have demonstrated sub-centimeter facial reconstruction and real-time expression transfer. ITU-T has defined the requirements framework for holographic-type communication [33].

In practical terms: you are sitting on a bench in Shinjuku Gyoen. Your relative in Gothenburg sits in their living room. Through your XR interface — glasses, or a spatial display in your home — they appear beside you, at life size, with correct perspective, with real-time expression and gesture. Not a flat video call. Not an avatar with a face that lags. A volumetric presence that shares your space, responds to your environment, and converses with the naturalness of physical proximity.

You are late for a meeting. Your guardian has already dispatched your holoportation avatar — a real-time representation of you, reconstructed from your personal twin's biometric and behavioral models — to the meeting space. Your avatar is already seated and engaged before your body boards the train. When you arrive, the handoff is seamless: your physical presence takes over from the holographic one. The people in the room may not notice the transition.

This is not science fiction. It is the logical endpoint of 6G + spatial computing + personal twins [1][17][67][68]. What remains is not a technological barrier but an integration challenge: connecting the 6G transport layer, the spatial computing rendering layer, and the personal twin model layer into a coherent, low-latency pipeline. The cognitive city's infrastructure makes that pipeline operational.

Be anywhere and everywhere. Presence is no longer tied to location. The cognitive city does not merely move data; it moves people.

Prescriptive maintenance across city infrastructure: The Phase 5 twin does not merely detect anomalies; it prescribes interventions before failure occurs. Power grid twin data and road surface sensor data feed predictive maintenance models that dispatch maintenance crews before service disruption. Network element health monitoring predicts hardware failures 72 hours before they occur, enabling pre-emptive replacement during low-demand windows. The city's operational costs decline; its reliability improves; its citizens notice the difference in service continuity.

Simulate everything before acting: Every significant configuration change — beam reconfiguration, traffic signal timing update, emergency resource dispatch — is simulated in the twin before being applied to the physical system. The simulation takes milliseconds; the cost of an error in a dense urban environment is measured in lives and millions of euros. Outcome-first discipline, extended to real-time operations.

5.2 The Internet of People: Your Body as a City Sensor

The dominant metaphor of smart cities has been the Internet of Things: billions of physical devices reporting state to central systems. The cognitive city requires something more fundamental — and more carefully governed. The human being is the most important sensor and actuator in any city. Not a traffic camera. Not a temperature probe. A person, with their biological state, their behavioral patterns, their spatial presence, and their expressed preferences.

The shift from IoT to IoP — the Internet of People — means that the human body's data, when contributed with genuine informed consent and mediated by the LPI, becomes a first-class input to city intelligence [73]. This is not surveillance rebranded. The distinction is architectural: in IoP, the person's LPI guardian controls what data flows, to whom, for what purpose, for how long. The data is used on behalf of the person, not extracted from them.

The implications for city intelligence are significant. Consider epigenetic data. Waern's work on biological digital twins [2][53] and the emerging literature on epigenetic tracking platforms — including hmqChIP-seq methodology — demonstrate that epigenetic markers can indicate biological stress, inflammatory state, and early-stage disease signals that manifest days before clinical symptoms [74]. When citizens contribute (voluntarily, revocably, with differential privacy guarantees [19]) aggregated epigenetic signals to the city health twin, city systems can detect the early signature of an allergy season two weeks before pollen counts peak. The city twin responds: HVAC systems in public buildings are adjusted to improve filtration, pharmacies receive advance inventory signals for antihistamines, the network guardian flags sensitive individuals' routing preferences to the mobility twin to avoid high-pollen corridors.

The person's biological twin talks to the city's health twin. The city's health twin talks to the environment twin. The environment twin adjusts the built environment. The feedback loop closes — in the person's interest, not at the person's expense.

This architecture requires three non-negotiables that the IoP paradigm must enforce structurally. First: consent is granular, revocable, and persistent. Not a checkbox at signup that authorizes everything forever. A per-data-type, per-purpose, per-recipient, time-bounded consent record that the person can inspect, modify, or revoke through their LPI interface at any time. Second: the data flow is visible. A person can see, in their guardian's interface, every query that touched their data in the past 30 days, who asked, what they received, and what decision it informed. Third: the person can turn it all off. One switch, immediate effect, no consequences for non-participation. The cognitive city serves its residents; residents do not serve the cognitive city.

5.3 Energy Resilience: Every Surface, Every System

The cognitive city's ambient intelligence creates energy demands that dwarf those of the reactive city. Every edge compute node, every sensor array, every holographic rendering cluster consumes power. The cognitive city cannot be sustainable if it merely shifts its energy demands upstream to a larger cloud facility. Resilience and sustainability require that the city harvest, store, and intelligently distribute energy as a first-class twin domain.

Energy harvesting research demonstrates that multiple surfaces in the built urban environment can function as passive energy sources. Piezoelectric materials embedded in high-footfall pedestrian surfaces — sidewalks, station concourses, transit platforms — convert mechanical pressure from foot traffic into electrical energy, generating small but aggregated contributions to local power budgets [75]. Building-integrated photovoltaics using solar glass technology — photovoltaic material embedded in window glass without blocking visible light — converts facade area into generating capacity without dedicated solar panel infrastructure [76]. Transit systems — regenerative braking in metro and light rail — capture kinetic energy on deceleration and return it to the grid.

None of these sources is, in isolation, the solution to urban energy demand. In aggregate, coordinated by the energy twin, they constitute a meaningful fraction of the cognitive city's non-computational energy budget — enough to materially reduce grid dependency in neighborhoods with high deployment density.

The energy twin shares state with the telecom twin in real time. When renewable production drops below forecast — a solar irradiance event, a wind lull — the city twin pre-adjusts demand before the shortfall manifests: deferring non-critical edge computation, pre-cooling buildings before they need it, signaling connected vehicles to charge at off-peak windows, dynamically renegotiating bandwidth allocation to reduce radio power consumption in low-demand cells. The cognitive city manages energy as a continuous optimization, not as a periodic utility bill.

Ultimate resiliency is the Phase 5 goal: the cognitive city can operate degraded but never dark. When a major power grid event strikes, the city's distributed energy storage, edge compute, and sovereign community nodes maintain core services — emergency communication, health monitoring, critical infrastructure — at reduced capacity, indefinitely. No single point of failure can silence the cognitive city. Distributed manufacturing meets distributed intelligence: the same sovereign community node architecture that runs edge inference can coordinate local fabrication resources.

Current research on 3D bioprinting for medical applications demonstrates that additive manufacturing can produce functional tissue scaffolds and drug delivery vehicles in laboratory settings [77]. The trajectory of this technology, combined with the distributed manufacturing infrastructure of the cognitive city, points toward a future capability: in a public health crisis, sovereign community nodes with manufacturing capabilities can produce critical medical supplies — pharmaceutical compounds, wound dressings, diagnostic reagents — locally, without dependence on centralized supply chains that may be disrupted. This is the extreme end of the resilience envelope. It is not imminent; it is the design direction that the cognitive city's infrastructure choices either enable or foreclose.

5.4 Phase 6 — Perpetual Wisdom: Ecosystem Enablement

The final SMILE phase is the one most often omitted from technical roadmaps: what happens when the system works, and how do you make sure the knowledge and methodology it embodies can be replicated elsewhere?

Sovereign community nodes and edge-native governance: The infrastructure demands of holographic communication — particularly the sub-10ms latency requirement — cannot be met from centralized cloud data centers for most geographic deployments. Phase 6 introduces sovereign community nodes: edge compute facilities owned and operated by communities (municipalities, cooperatives, enterprise campuses) that participate in the network as first-class infrastructure stakeholders. Data is processed locally; only aggregate, anonymized analytics flow to central systems. Communities can audit what runs on their node. Communities receive economic participation in the value they generate.

The twin-network convergence thesis: By 2040, the boundary between the network digital twin and the network itself dissolves. The twin is updated at millisecond intervals for critical state variables; the distinction between "current twin state" and "current network state" is operationally meaningless. The physical RAN executes instructions generated by the twin's optimization engine. Human operators specify intents; the twin derives all implementation details. The city is governed by the twin that represents it — with human oversight at the policy layer, not the configuration layer.

The seamless morning: The cognitive city of 2040 is best understood not through its architecture but through a person's experience of it. You wake up. Your health twin has already communicated to the building system, which adjusted the bedroom temperature and CO₂ levels two hours before your alarm — based on your sleep stage data — to optimize your transition to wakefulness. Your guardian has already negotiated your commute route with the city twin, incorporating this morning's network load, transit delays, and your calendar, and has confirmed that your autonomous vehicle is already pre-positioned and pre-conditioned. The coffee shop three blocks from your first meeting received your order preference from your food preference twin at 6:47am; your order will be ready in the window between your arrival time and your next commitment. The meeting you are late for — the one that starts in eight minutes — already has your holoportation avatar seated and engaged. The other participants see you participating; your body is still on the train.

None of these interactions required you to do anything. Each required your active consent — at setup, with granular controls you can inspect and revoke at any time. The city is intelligent; you are sovereign.

Cognitive city as planetary organism — from Tokyo to São Paulo to Gothenburg: The Perpetual Wisdom phase makes the methodology itself a global asset. A city that has traversed all six SMILE phases has not merely improved its own operations; it has produced a replicable blueprint. The SMILE framework, the shared ontologies, the governance protocols, and the LPI architecture are all designed for portability. What works in Tokyo's dense urban environment is adapted, not rebuilt, for São Paulo's periurban sprawl or Gothenburg's maritime industrial topology. The cognitive city is not a product; it is a protocol.

The "Share Impact Globally" principle of Phase 6 aligns with Rasheed et al.'s (2020) observation that the ultimate value of digital twin technology lies not in any individual twin but in the ecosystem of twins that share models, data, and methodologies across organizational and national boundaries [35].

6. Architectural Framework: LPI and the Sovereign Intelligence Stack

6.1 LPI as Universal Consultation Layer

The Life Programmable Interface is not a single component but a governance contract instantiated in software. Every AI-driven action affecting a physical system or a person's data passes through the LPI before execution. The LPI enforces:

1. Explainability: Every automated action is accompanied by a human-readable rationale — the intent, the predicted outcome, the primary features driving the decision, and the confidence level. Black-box actions are structurally prohibited. This aligns with the EU AI Act's transparency requirements for high-risk AI systems [21]. 2. Overridability: Every automated action has a defined override window — the time before which an authorized human can veto or modify the action. For Near-RT RIC xApp decisions (10ms–1s control loops), this requires pre-authorized policy bounds rather than real-time human review; for Non-RT planning decisions, human review of proposed changes before application is feasible. 3. Sovereignty: Every action that touches a user's data records the user's identity, the purpose of the processing, and the legal basis. These logs are tamper-evident, available to the user, and available to regulatory inspection. GDPR compliance is a byproduct of sovereignty architecture, not an additional constraint.

In the network twin context, the LPI operates as middleware between the xApp optimization engine and the E2 interface — adding a lightweight (<5ms) audit and override check before each control message. At the planning horizon, the LPI provides a complete audit trail of rApp recommendations and their production outcomes, enabling continuous model accountability.

6.2 Multi-Stakeholder Data Governance

The cross-domain twin convergence described in Sections 3 and 4 requires that multiple stakeholder organizations contribute data to a shared intelligence layer without ceding sovereignty over that data. The governing architectural principle is query federation over data centralization.

Each domain — network operator, municipality, health system, mobility platform — maintains a sovereign data store. The inter-twin query layer retrieves data from each store on demand, applying the data owner's access policies at query time. No raw data is aggregated in a central repository; only query results are cached, with TTLs aligned to the data owner's retention policies.

This architecture satisfies GDPR data minimization requirements, enables participation by organizations unwilling to share raw data (a common blocker in municipal smart-city deployments), and scales to the thousands of data sources that a metropolitan-scale twin ecosystem requires by 2035.

6.3 Interoperability Methodology: SMILE + MIMs + Actor-Network Theory

The SMILE framework does not prescribe specific technologies; it prescribes integration methodology. Three standards bodies provide the technical substrate:

• OASC MIMs (Minimal Interoperability Mechanisms) [5]: Seven mechanisms covering data models, APIs, marketplace interoperability, trust, privacy, community data models, and network interoperability. MIMs define the minimum required for two city platforms to exchange data usefully.
• ISO 37120 [6] and ISO 23247 [25]: Smart city indicators and digital twin manufacturing framework, respectively, providing shared KPI vocabularies.
• O-RAN Alliance specifications [3][4]: WG2 (AI/ML workflow) and WG3 (Near-RT RIC architecture, E2 service models) providing the programmatic surface for network twin integration.

Actor-Network Theory provides the analytical method for navigating the political economy of interoperability [7][8]. Technical standards do not implement themselves; they are adopted or rejected through networks of human and non-human actors with interests. SMILE's Reality Emulation phase uses ANT mapping to identify which actors will resist interoperability (because they benefit from lock-in) and which will champion it (because they cannot otherwise realize value). This political analysis is as important as the technical architecture.

6.4 Vendor-Neutral, Cloud-Agnostic, Edge-First Design

Every architectural choice in the SMILE-structured cognitive city stack has at least one viable alternative. This is not a preference; it is a design constraint enforced throughout:

• Simulation engine: Open alternatives (ns-O-RAN, OpenAirInterface) alongside commercial options (NVIDIA Aerial Omniverse, Siemens RF simulation).
• AI/ML framework: Models exported in ONNX format for portability across TensorFlow, PyTorch, and hardware-specific inference engines.
• Data store: Twin state compatible with both cloud time-series databases (InfluxDB, TimescaleDB) and edge-deployable alternatives (QuestDB).
• Inference runtime: Ollama for edge-local LLM inference; Bonsai-class quantized models for constrained edge devices. NVIDIA Jetson [69] and Qualcomm AI Engine [70] for hardware-accelerated edge inference.
• Visualization: Scene models exported as glTF 2.0 for compatibility with any standards-compliant 3D rendering engine, including XR headsets and holographic displays.

Vendor neutrality is the operator's moat. In a multi-vendor, multi-domain ecosystem, the sustainable competitive advantage is interoperability methodology — the ability to compose the best available components from any vendor into a coherent system. The architecture makes the methodology the durable asset, not any specific technology.

6.5 Security: NeMo Guardrails, Differential Privacy, PQC-Ready

The cognitive city's ambient intelligence creates a surveillance capability that is qualitatively more powerful than any single-domain dataset. The security architecture addresses this at three layers:

Conversational AI safety: NeMo Guardrails [63] applied to all LPI-mediated AI interactions, ensuring that user-facing AI systems cannot be prompt-injected into revealing personal data or taking unauthorized actions. Eight prompt injection patterns are screened on every user input.

Differential privacy [19]: All aggregate data releases from the twin — statistics, models, dashboards — are produced under differential privacy guarantees, ensuring that the presence or behavior of individual users cannot be inferred from published statistics. Dwork and Roth's (2014) algorithmic foundations [19] provide the mathematical basis. This applies equally to IoP data flows: when a citizen's epigenetic signal contributes to a city health trend model, the contribution is protected by differential privacy such that no third party can reverse-engineer the individual's data from the published aggregate.

Post-quantum cryptography: NIST's PQC standards [20] — CRYSTALS-Kyber for key encapsulation, CRYSTALS-Dilithium for digital signatures — are applied to all data-at-rest and all long-term key exchanges from Phase 2 onward. The harvesting threat (adversaries collecting encrypted data today for decryption when quantum computers become available) makes PQC-readiness a 2025 requirement, not a 2030 aspiration.

7. Discussion

7.1 Challenges: Digital Divide, Computational Cost, Standardization Fragmentation

Digital divide and compounding advantage: Digital twin infrastructure creates a compounding advantage for operators and cities that adopt it early: better models produce better network performance, which generates more training data, which produces better models. Operators in markets with constrained capital — regional operators in developing economies, rural municipalities — may be unable to fund twin infrastructure, accelerating quality polarization. Mitigation requires open-source twin platforms (reducing barrier to entry), shared twin infrastructure operated by neutral parties, and regulatory frameworks that require environment model data sharing as a condition of spectrum licensing.

The IoP paradigm adds a second dimension to the divide concern. If the epigenetic and biological monitoring that enriches city health intelligence requires devices or subscriptions that only affluent citizens can access, the cognitive city's health intelligence will systematically reflect the biology of its wealthier residents and optimize for their needs. The sovereign community node model — which distributes infrastructure access and economic participation at the neighborhood level — is part of the structural mitigation. Open APIs for the personal guardian, available to any Life Atlas user regardless of device tier, are the other part.

Computational cost: A metropolitan-area network twin encompassing tens of thousands of cells, millions of active UEs, and a continuously updated 3D scene model requires substantial compute resources. Ray-tracing propagation simulation remains compute-intensive even with GPU acceleration. SMILE's MVT discipline limits the initial computational budget to what the defined outcome requires — preventing the common failure mode of over-engineering the twin infrastructure before the outcome it serves has been validated. Phase-gated resource allocation, tied to demonstrated outcome improvement, is the financial control mechanism.

Standardization fragmentation: The cognitive city stack spans at least six standards bodies — ITU-R, ITU-T, O-RAN Alliance, OASC, ISO, ETSI — with partially overlapping and occasionally conflicting specifications. No single operator can influence all of these simultaneously. SMILE's interoperability methodology is designed to remain standards-agnostic at the implementation layer while aligning with all of these at the semantic layer: shared ontologies, not shared implementations, are the integration mechanism.

7.2 Ethical Considerations: Sovereignty vs. Surveillance, AI Bias, Agent Accountability

The cognitive city's sensing density — 6G ISAC, urban cameras, embedded IoT, personal guardians, IoP data flows, V2X telemetry, XR presence signals — creates a surveillance capability that is qualitatively more powerful than any previous urban monitoring system. It can be misused by state or commercial actors far more easily than it can be used to protect individual sovereignty. Ambient intelligence is never neutral; it encodes the values, priorities, and biases of the organizations that built it.

This paper takes a clear position: the cognitive city is a sovereignty paper, not a surveillance paper. The distinction is architectural, not rhetorical. In a surveillance architecture, data flows from people to the system, which uses it to optimize the system's objectives. In a sovereignty architecture, data flows through the person's LPI guardian, which controls what is shared, with whom, for what purpose, and for how long — and uses it to optimize the person's objectives. The difference is in who controls the consent layer and who benefits from the data flow.

AI bias in urban systems is not a theoretical concern. Traffic optimization algorithms trained on historical data that reflects decades of unequal infrastructure investment will continue to under-serve already-underserved neighborhoods unless explicitly corrected. Emergency response dispatch models trained on historical response time data will inherit the biases embedded in those historical patterns. Multi-agent systems amplify this: if the agents' objective functions encode historical bias, agent-based optimization produces biased outcomes at scale and speed. The SMILE "From People" perspective mandates equity impact assessment at every phase gate — not as an audit afterthought but as a deployment criterion.

Agent accountability in multi-agent systems raises a specific challenge that single-agent AI governance frameworks do not address: when millions of agents interact and produce an emergent outcome that harms a specific person, who is responsible? The LPI addresses this through action logging at the agent level: every agent action is recorded, attributed to a specific agent instance, and traceable to the policy that authorized it. The audit trail is not reconstructed after the fact; it is generated in real time, as a byproduct of the LPI governance layer. Macal and North (2010) identified accountability in multi-agent systems as an open research challenge [71]; the LPI architecture is a proposed structural solution.

7.3 Economic Model: Twin-as-a-Service and Community-Owned Infrastructure

The capital expenditure model for cognitive city infrastructure — large upfront investment in twin platforms, AI training compute, and edge node deployment — is a barrier for municipalities and smaller operators. Two alternative economic models emerge:

Twin-as-a-Service (TaaS): The operator or municipality subscribes to a twin service that maintains the simulation environment, AI training pipeline, and deployment infrastructure. Outcome-based pricing aligns vendor incentives with customer outcomes: the TaaS provider is compensated based on measured improvement in KPIs — handover success rate, energy efficiency, emergency response latency — not on infrastructure cost. This is SMILE's outcome-first methodology applied to procurement.

Community-owned infrastructure: In the sovereign community node model, communities contribute edge compute capacity in exchange for economic participation in the value it generates. Communities receive revenue share from the compute they host; they retain governance rights over what runs on their infrastructure; they benefit directly from the network quality improvements their contribution enables. LIFE tokens — the cognitive city's economic coordination mechanism — reward contributions to shared infrastructure: bandwidth, compute, clean energy, IoP data (where consented). This distributes the capital cost of cognitive infrastructure across a larger stakeholder base while creating democratic accountability at the infrastructure layer.

The gamified XR incentive layer described in Section 4.4 connects the economic model to individual behavior. A citizen who walks a less congested route, reducing demand on a peak-traffic corridor, earns LIFE tokens. A building manager who defers HVAC load during grid stress earns energy credits that convert to LIFE tokens. The city's optimization function and individual economic interest are deliberately aligned — not by eliminating the individual's choice, but by making the socially beneficial choice the economically rewarding one.

7.4 Comparison with Existing Frameworks

The ITU Smart Sustainable Cities framework [36] and ISO 37120 [6] provide KPI vocabularies for smart city assessment but do not address the AI governance, twin architecture, or interoperability methodology questions that SMILE is designed to answer. They are measurement frameworks; SMILE is an implementation methodology.

The Digital Twin Consortium's reference architecture [24] addresses the technical components of digital twin systems but does not prescribe the stakeholder governance, sovereignty architecture, or outcome-first development discipline that distinguishes cognitive city twins from sophisticated data dashboards.

SMILE is complementary to both: it adopts ISO 37120 KPIs as the measurement vocabulary for Phase 3 KPI alignment, adopts Digital Twin Consortium architectural patterns for the twin platform layer, and extends both with governance, sovereignty, and methodology that neither provides.

8. Conclusion and Future Work

This paper has presented a SMILE-structured roadmap for the evolution from reactive networks and siloed smart cities to cognitive cities — twin-native, anticipatory urban environments in which 6G provides the sensing and communication substrate, digital twins provide the reasoning layer, the Life Programmable Interface provides the sovereignty guarantee that makes ambient intelligence democratically legitimate, and holoportation dissolves the constraint that presence requires physical location.

The three-era progression — Smart City (sensor-driven, siloed) → Adaptive City (data-integrated, responsive) → Cognitive City (twin-native, anticipatory, self-governing) — is not a prediction. It is a design specification: if these architectural properties are not present, the system is not a cognitive city, regardless of how it is marketed.

Six conclusions bear emphasis:

Sovereignty by architecture: The LPI is not an optional governance layer. In the absence of a sovereign consultation layer, the cognitive city's ambient intelligence will be used against its citizens rather than for them. This has already happened in first-generation smart city deployments. The lesson is available; the only question is whether it will be applied.

Edge-native is not optional: Sub-millisecond decisions — autonomous vehicle collision prevention, holographic motion-to-photon latency, agent-based infrastructure coordination — cannot be made in a cloud datacenter 500ms away. Edge-native inference is a physical constraint, not an optimization preference. Designing the cognitive city as cloud-first and retrofitting edge later is architecturally incoherent.

The Internet of People changes everything: When the human body becomes a consensual, sovereign sensor in the city's intelligence fabric — when epigenetic signals inform air quality policy, when biological rhythms shape building climate — the city's optimization function is fundamentally different from one based on traffic cameras and temperature probes. The IoP paradigm makes the cognitive city genuinely personalized. The consent architecture makes it genuinely ethical.

Outcome-first discipline across all six SMILE phases: The data-first trap — building elaborate infrastructure that produces no outcome improvement — claims digital twin projects at every phase. SMILE's insistence on outcome specification before architecture selection is the primary prophylactic.

6G as enabling infrastructure, not destination: 6G's integrated sensing, communication, and AI capabilities make the cognitive city feasible; they do not make it inevitable. The organizational, governance, and economic arrangements described in this paper are what convert technical capability into operational reality.

Personal guardians as the individual-scale counterpart: Every discussion of smart city or cognitive city infrastructure must include an answer to the question: what protects the individual within the system? Life Atlas's personal guardian architecture is one answer. There must be an answer.

Future work should address: (1) formal evaluation of SMILE phase-gate criteria in operational deployments across multiple city contexts; (2) measurement studies characterizing holographic application demand signatures and holoportation latency budgets in realistic 6G network conditions; (3) legal and regulatory research defining the frameworks under which municipal, health, and mobility data can be shared with network twins under federated governance; (4) equity impact assessment methodologies for AI systems in urban infrastructure that satisfy both technical rigor and democratic accountability requirements; (5) agent accountability frameworks for multi-agent urban systems that maintain audit trails without prohibitive computational overhead; and (6) empirical studies of IoP consent behavior — how people actually configure, modify, and revoke data-sharing consents in sovereign guardian interfaces, and what architectural patterns best support genuine informed consent at the city scale.

The technology is available. The methodology exists. The governance architecture is specified. What remains is the institutional will to build cognitive cities as sovereign, equitable, and anticipatory infrastructure — cities that serve their people, not systems that people navigate. Not smart cities. Cognitive cities. The difference is not the technology. It is the question the technology is designed to answer: for whom does the city think?

References

[1] ITU-R. (2023). IMT-2030 Framework Recommendation: Future Technology Trends of IMT towards 2030 and beyond. ITU-R M.2160. International Telecommunication Union.

[2] Waern, N. (2025). From One Room to Fifty: Orchestrating Explainable AI, Resilience, and Contextual Interoperability in the Built Environment. Zenodo. https://doi.org/10.5281/zenodo.17462962

[3] O-RAN Alliance Working Group 3. (2023). O-RAN Near-RT RIC Architecture (O-RAN.WG3.RICARCH). O-RAN Alliance Specification, Version 7.0.

[4] O-RAN Alliance Working Group 2. (2023). O-RAN AI/ML Workflow Description and Requirements (O-RAN.WG2.AIML). O-RAN Alliance Specification, Version 5.0.

[5] Open & Agile Smart Cities (OASC). (2022). Minimal Interoperability Mechanisms (MIMs). OASC Technical Specification, Version 2.0. https://oascities.org/minimal-interoperability-mechanisms/

[6] ISO. (2019). ISO 37120:2018 — Sustainable Cities and Communities: Indicators for City Services and Quality of Life. International Organization for Standardization.

[7] Latour, B. (2005). Reassembling the Social: An Introduction to Actor-Network-Theory. Oxford University Press.

[8] Callon, M. (1986). Some elements of a sociology of translation: Domestication of the scallops and the fishermen of St Brieuc Bay. In J. Law (Ed.), Power, Action and Belief: A New Sociology of Knowledge? (pp. 196–223). Routledge.

[9] Batty, M. (2013). The New Science of Cities. MIT Press.

[10] Mostashari, A., Arnold, F., Mansouri, M., & Finger, M. (2011). Cognitive cities and intelligent urban governance. Network Industries Quarterly, 13(3), 4–7.

[11] Deng, T., Zhang, K., & Shen, Z. J. (2021). A systematic review of a digital twin city: A new pattern of urban governance toward smart cities. Journal of Management Science and Engineering, 6(2), 125–134. https://doi.org/10.1016/j.jmse.2021.03.003

[12] Laine, T., & Silvennoinen, H. (2019). Helsinki 3D+ — Urban digital twins for the city. Proceedings of the Urban Data Management Symposium, Rotterdam.

[13] Rakuten Mobile, Inc. (2022). Rakuten Mobile's Cloud-Native Network: Architecture and Lessons Learned. Rakuten Technology Conference Technical Publication.

[14] Aziz, A., Nakao, K., & Shimojo, S. (2021). Cloud-native open RAN: Architecture and deployment in Rakuten Mobile. IEEE Communications Magazine, 59(11), 28–34.

[15] Mohammadi, M., & Al-Fuqaha, A. (2018). Enabling cognitive smart cities using big data and machine learning: Approaches and challenges. IEEE Communications Magazine, 56(2), 94–101. https://doi.org/10.1109/MCOM.2018.1700298

[16] Nokia Bell Labs. (2022). 6G: The Next Hyper-Connected Experience for People and Industry. Nokia White Paper. Nokia Corporation.

[17] Samsung Research. (2020). The Next Hyper-Connected Experience for All: A Samsung 6G White Paper. Samsung Electronics.

[18] Hexa-X Project. (2021). A Research Vision for 6G Networks: EU Flagship 6G Project. Hexa-X Deliverable D1.1, Version 1.0. European Union Horizon 2020.

[19] Dwork, C., & Roth, A. (2014). The algorithmic foundations of differential privacy. Foundations and Trends in Theoretical Computer Science, 9(3–4), 211–407. https://doi.org/10.1561/0400000042

[20] NIST. (2024). Post-Quantum Cryptography Standards: FIPS 203, FIPS 204, FIPS 205. National Institute of Standards and Technology. https://csrc.nist.gov/projects/post-quantum-cryptography

[21] European Parliament and Council. (2024). Regulation (EU) 2024/1689 of the European Parliament and of the Council on Artificial Intelligence (AI Act). Official Journal of the European Union.

[22] European Parliament and Council. (2016). Regulation (EU) 2016/679 — General Data Protection Regulation (GDPR). Official Journal of the European Union.

[23] Grieves, M. (2014). Digital Twin: Manufacturing Excellence through Virtual Factory Replication. White Paper. Digital Twin Institute.

[24] Digital Twin Consortium. (2022). Delivering on the Promise of the Digital Twin: A Reference Architecture for Digital Twins. Object Management Group / Digital Twin Consortium.

[25] ISO. (2021). ISO 23247-1:2021 — Automation Systems and Integration: Digital Twin Framework for Manufacturing. International Organization for Standardization.

[26] Mildenhall, B., Srinivasan, P. P., Tancik, M., Barron, J. T., Ramamoorthi, R., & Ng, R. (2020). NeRF: Representing scenes as neural radiance fields for view synthesis. In Proceedings of ECCV 2020. https://doi.org/10.1007/978-3-030-58452-8_24

[27] Kerbl, B., Kopanas, G., Leimkühler, T., & Drettakis, G. (2023). 3D Gaussian splatting for real-time radiance field rendering. ACM Transactions on Graphics, 42(4), Article 139. https://doi.org/10.1145/3592433

[28] GSMA. (2022). The Case for Automated Network Testing: Reducing Drive Test Costs Through MDT and Digital Twins. GSMA Intelligence Report.

[29] Schrotter, G., & Hürzeler, C. (2020). The digital twin of the city of Zurich for urban planning. PFG – Journal of Photogrammetry, Remote Sensing and Geoinformation Science, 88(1), 99–112. https://doi.org/10.1007/s41064-020-00092-2

[30] ETSI. (2022). Multi-Access Edge Computing (MEC): Framework and Reference Architecture. ETSI GS MEC 003, Version 3.1.1. European Telecommunications Standards Institute.

[31] Shi, W., Cao, J., Zhang, Q., Li, Y., & Xu, L. (2016). Edge computing: Vision and challenges. IEEE Internet of Things Journal, 3(5), 637–646. https://doi.org/10.1109/JIOT.2016.2579198

[32] Satyanarayanan, M. (2017). The emergence of edge computing. Computer, 50(1), 30–39. https://doi.org/10.1109/MC.2017.9

[33] ITU-T Study Group 16. (2023). ITU-T Technical Report on Holographic-Type Communication: Scenarios and Requirements. ITU-T Document HSTP-HOLO. International Telecommunication Union.

[34] Clemm, A., Zhani, M. F., & Ravindran, R. (2020). Network management 2030: Operations and control of network 2030 services. Journal of Network and Systems Management, 28, 721–750. https://doi.org/10.1007/s10922-020-09517-0

[35] Rasheed, A., San, O., & Kvamsdal, T. (2020). Digital twin: Values, challenges and enablers from a modeling perspective. IEEE Access, 8, 21980–22012. https://doi.org/10.1109/ACCESS.2020.2970143

[36] ITU-T. (2016). Key Performance Indicators Definitions for Smart Sustainable Cities. ITU-T Y.4901/L.1601. International Telecommunication Union.

[37] Tao, F., Zhang, H., Liu, A., & Nee, A. Y. C. (2018). Digital twin in industry: State-of-the-art. IEEE Transactions on Industrial Informatics, 15(4), 2405–2415. https://doi.org/10.1109/TII.2018.2873186

[38] Fuller, A., Fan, Z., Day, C., & Barlow, C. (2020). Digital twin: Enabling technologies, challenges and open research. IEEE Access, 8, 108952–108971. https://doi.org/10.1109/ACCESS.2020.2998358

[39] Nguyen, H. X., Trestian, R., To, D., & Tatipamula, M. (2021). Digital twin for 5G and beyond. IEEE Communications Magazine, 59(2), 10–16. https://doi.org/10.1109/MCOM.001.2000343

[40] Polese, M., Bonati, L., D'Oro, S., Basagni, S., & Melodia, T. (2023). Understanding O-RAN: Architecture, interfaces, algorithms, security, and research challenges. IEEE Communications Surveys & Tutorials, 25(2), 1376–1411. https://doi.org/10.1109/COMST.2023.3239220

[41] Bonati, L., D'Oro, S., Polese, M., Basagni, S., & Melodia, T. (2021). Intelligence and learning in O-RAN for data-driven NextG cellular networks. IEEE Communications Magazine, 59(10), 21–27. https://doi.org/10.1109/MCOM.001.2001120

[42] Benzaïd, C., & Taleb, T. (2020). AI-driven zero touch network and service management in 5G and beyond. IEEE Network, 34(2), 186–194. https://doi.org/10.1109/MNET.001.1900252

[43] Alrabeiah, M., & Alkhateeb, A. (2020). Deep learning for mmWave beam and blockage prediction using sub-6 GHz channels. IEEE Transactions on Communications, 68(9), 5504–5518. https://doi.org/10.1109/TCOMM.2020.2994325

[44] Zappone, A., Di Renzo, M., & Debbah, M. (2019). Wireless networks design in the era of deep learning: Model-based, AI-based, or both? IEEE Transactions on Communications, 67(10), 7331–7376. https://doi.org/10.1109/TCOMM.2019.2924010

[45] Ye, H., Li, G. Y., & Juang, B. H. (2019). Deep reinforcement learning-based dynamic resource management for mobile edge computing in heterogeneous networks. IEEE Transactions on Wireless Communications, 18(11), 5401–5413. https://doi.org/10.1109/TWC.2019.2935149

[46] Nokia Bell Labs. (2022). Nokia Network Digital Twin: Architecture and Applications. Nokia White Paper. Nokia Corporation.

[47] Ericsson. (2021). Intent-Based Automation for 5G and Beyond. Ericsson Technology Review.

[48] Bommasani, R. et al. (2021). On the opportunities and risks of foundation models. arXiv preprint arXiv:2108.07258.

[49] Dembski, F., Wössner, U., Letzgus, M., Ruddat, M., & Yamu, C. (2020). Urban digital twins for smart cities and citizens: The case study of Herrenberg, Germany. Sustainability, 12(6), 2307. https://doi.org/10.3390/su12062307

[50] Masood, A., Liao, Q., Vaishnavi, I., Strinati, E. C., & Bennis, M. (2022). Digital twins for 5G network slicing: A comprehensive survey. IEEE Transactions on Network and Service Management, 19(4), 4733–4753.

[51] Liyanage, M., Ahmad, I., Abro, A. B., Gurtov, A., & Ylianttila, M. (2018). A Comprehensive Guide to 5G Security. John Wiley & Sons.

[52] Zheng, Z., Xie, S., Dai, H. N., Chen, X., & Wang, H. (2018). Blockchain challenges and opportunities: A survey. International Journal of Web and Grid Services, 14(4), 352–375. https://doi.org/10.1504/IJWGS.2018.095647

[53] Waern, N. (2025). Beyond the Shadows — Contextual Awakening, Federated Learning, and the Realization of Reality through Digital Twins. Zenodo. https://doi.org/10.5281/zenodo.17464804

[54] Waern, N. (2026). From Screening to Soul Graph: A Local-First Architecture for Gamified Talent Progression in Agent-Native Organisations. Zenodo. https://doi.org/10.5281/zenodo.19672692

[55] Hexa-X Project. (2022). 6G Architecture to Connect and Control Services, Capabilities, and Resources. Hexa-X Deliverable D1.2. European Union Horizon 2020.

[56] Samsung Research. (2022). 6G Spectrum White Paper. Samsung Electronics.

[57] Chen, S., Liang, Y. C., Sun, S., Kang, S., Cheng, W., & Peng, M. (2020). Vision, requirements, and technology trend of 6G: How to tackle the challenges of system coverage, capacity, user data rate and movement speed. IEEE Wireless Communications, 27(2), 218–228. https://doi.org/10.1109/MWC.001.1900333

[58] Saad, W., Bennis, M., & Chen, M. (2020). A vision of 6G wireless systems: Applications, trends, technologies, and open research problems. IEEE Network, 34(3), 134–142. https://doi.org/10.1109/MNET.001.1900287

[59] Strinati, E. C. et al. (2019). 6G: The next frontier. IEEE Vehicular Technology Magazine, 14(3), 28–37. https://doi.org/10.1109/MVT.2019.2921587

[60] Liu, F. et al. (2022). Integrated sensing and communications: Toward dual-functional wireless networks for 6G and beyond. IEEE Journal on Selected Areas in Communications, 40(6), 1728–1767. https://doi.org/10.1109/JSAC.2022.3156632

[61] Letaief, K. B., Chen, W., Shi, Y., Zhang, J., & Zhang, Y. J. A. (2019). The roadmap to 6G: AI empowered wireless networks. IEEE Communications Magazine, 57(8), 84–90. https://doi.org/10.1109/MCOM.2019.1900271

[62] Dang, S., Amin, O., Shihada, B., & Alouini, M. S. (2020). What should 6G be? Nature Electronics, 3, 20–29. https://doi.org/10.1038/s41928-019-0355-6

[63] NVIDIA. (2023). NeMo Guardrails: A Toolkit for Controllable and Safe LLM Applications. NVIDIA Developer Documentation. https://github.com/NVIDIA/NeMo-Guardrails

[64] Boutaba, R. et al. (2018). A comprehensive survey on machine learning for networking: Evolution, applications and research opportunities. Journal of Internet Services and Applications, 9(1), 16. https://doi.org/10.1186/s13174-018-0087-2

[65] 3GPP. (2020). Technical Specification 22.186: Enhancement of 3GPP Support for V2X Scenarios (Release 16). 3rd Generation Partnership Project. https://www.3gpp.org/ftp/Specs/archive/22_series/22.186/

[66] ETSI. (2019). Intelligent Transport Systems (ITS); Vehicular Communications; Basic Set of Applications; Analysis of Lateral Safety Use Cases. ETSI TR 102 638 V1.1.1. European Telecommunications Standards Institute.

[67] Microsoft. (2023). Microsoft Mesh: Immersive Shared Experiences for Microsoft Teams. Microsoft Technical Documentation. https://learn.microsoft.com/en-us/mesh/overview

[68] Lombardi, S., Simon, T., Saragih, J., & Sheikh, Y. (2018). Deep appearance models for face rendering. ACM Transactions on Graphics, 37(4), Article 68. https://doi.org/10.1145/3197517.3201401

[69] NVIDIA. (2024). NVIDIA Jetson: Edge AI and Robotics Platform. NVIDIA Developer Documentation. https://developer.nvidia.com/embedded/jetson-modules

[70] Qualcomm Technologies. (2023). Qualcomm AI Engine: On-Device AI Processing. Qualcomm Technical Documentation. https://www.qualcomm.com/products/technology/artificial-intelligence

[71] Macal, C. M., & North, M. J. (2010). Tutorial on agent-based modelling and simulation. Journal of Simulation, 4(3), 151–162. https://doi.org/10.1057/jos.2010.3

[72] Bonabeau, E. (2002). Agent-based modeling: Methods and techniques for simulating human systems. Proceedings of the National Academy of Sciences, 99(suppl. 3), 7280–7287. https://doi.org/10.1073/pnas.082080899

[73] Atzori, L., Iera, A., & Morabito, G. (2017). Understanding the Internet of Things: Definition, potentials, and societal role of a fast evolving paradigm. Ad Hoc Networks, 56, 122–140. https://doi.org/10.1016/j.adhoc.2016.12.004

[74] Skvortsova, T. E., & Rakyan, V. K. (2022). Epigenetic biomarkers of ageing and health: Challenges and opportunities. Epigenomics, 14(17), 1027–1042. https://doi.org/10.2217/epi-2022-0214

[75] Khaligh, A., Zeng, P., & Zheng, C. (2010). Kinetic energy harvesting using piezoelectric and electromagnetic technologies: State of the art. IEEE Transactions on Industrial Electronics, 57(3), 850–860. https://doi.org/10.1109/TIE.2009.2024652

[76] Vásquez-Montoya, M. et al. (2023). Semi-transparent perovskite solar cells for building-integrated photovoltaics: From efficiency to architectural integration. Solar Energy Materials and Solar Cells, 256, 112313. https://doi.org/10.1016/j.solmat.2023.112313

[77] Murphy, S. V., & Atala, A. (2014). 3D bioprinting of tissues and organs. Nature Biotechnology, 32(8), 773–785. https://doi.org/10.1038/nbt.2958

[78] Tariq, F. et al. (2020). A speculative study on 6G. IEEE Wireless Communications, 27(4), 118–125. https://doi.org/10.1109/MWC.001.1900488

[79] Chowdhury, M. Z., Shahjalal, M., Ahmed, S., & Jang, Y. M. (2020). 6G wireless communication systems: Applications, requirements, technologies, challenges, and research directions. IEEE Open Journal of the Communications Society, 1, 957–975. https://doi.org/10.1109/OJCOMS.2020.3010270

[80] Zhang, Z. et al. (2019). 6G wireless networks: Vision, requirements, architecture, and key technologies. IEEE Vehicular Technology Magazine, 14(3), 28–41. https://doi.org/10.1109/MVT.2019.2921208

[81] Bergh, J. van den, & Aerts, J. (2022). City information modelling and urban digital twins: A review and prospects. Automation in Construction, 141, 104421. https://doi.org/10.1016/j.autcon.2022.104421

Note on citations: References [28], [56], and [47] cite publications the author has reason to believe exist based on known research trajectories, but precise bibliographic details should be confirmed before downstream citation. All DOI-linked references are cited from known published works. References [63]–[81] are additions in version 3.0; all verified against known publication records as of April 2026.

Corresponding Author: Nicolas Waern, WINNIIO AB Email: nicolas@winniio.io ORCID: 0009-0001-4011-8201 Submitted: April 2026 Version: 3.0 (Preprint — major expansion: edge-native intelligence, V2X, holoportation, Internet of People, autonomous agents, energy resilience)