15. Emerging Networks and Research Directions

Introduction and Context

Computer networks are entering a period in which connectivity is no longer defined only by faster links, larger routers, or higher capacity data centres. Networks are becoming heterogeneous, software-defined, AI-assisted, satellite-enabled, security-sensitive, and increasingly responsible for managing physical and societal infrastructure. The emerging network landscape combines terrestrial 5G and prospective 6G systems, low-Earth-orbit satellite constellations, intelligent control loops, quantum technologies, information-centric communication, and energy-aware operation.

These developments should not be treated as isolated technologies. They represent different responses to persistent networking problems: limited spectrum, uneven coverage, complex service requirements, dependence on centralized infrastructure, rising cyber risk, excessive energy consumption, and the difficulty of managing highly dynamic systems. Some proposals are mature enough for deployment, whereas others remain research visions whose feasibility depends on hardware, regulation, economics, and standardization.

A postgraduate study of emerging networks therefore requires critical evaluation rather than technology promotion. The relevant questions are: What problem does a proposed architecture solve? Which assumptions does it make? How can its performance be measured experimentally? What new risks does it introduce? How does it interact with existing Internet protocols and operational practices? This chapter examines these questions and develops a research-oriented framework for evaluating emerging network directions.

Learning Objectives

After studying this chapter, the learner should be able to:

1. Explain the architectural and service objectives associated with 6G and integrated terrestrial–non-terrestrial networks.

2. Distinguish AI-assisted, AI-native, and autonomous networking.

3. Describe the principles and practical constraints of quantum networking and post-quantum security.

4. Analyse information-centric networking and green networking as alternative responses to current Internet limitations.

5. Design an experimental methodology for comparing emerging network architectures.

6. Critically assess performance, security, governance, energy, and deployment trade-offs.

7. Apply these ideas to realistic organizational and infrastructure scenarios.

From Faster Connectivity to Networked Intelligence

Earlier generations of mobile communication were commonly characterized by a dominant performance objective: voice quality, mobile broadband, or improved data rates. Emerging networks instead support diverse service classes. A factory may require deterministic communication for robotic control, a remote region may require intermittent but wide-area satellite access, and an immersive application may require high capacity with tightly bounded latency and jitter. These demands cannot be satisfied by a single uniform network design.

A useful way to reason about an emerging network is through five interacting planes:

• Data plane: forwarding, radio transmission, routing, caching, and packet processing.
• Control plane: path selection, mobility management, resource allocation, and policy enforcement.
• Knowledge plane: telemetry, models, context, predictions, and digital representations of the network.
• Trust plane: identity, authorization, cryptographic protection, attestation, and auditability.
• Sustainability plane: energy, carbon intensity, equipment lifetime, and resource efficiency.

Traditional architectures emphasize the first two planes. Emerging networks increasingly connect all five. This integration creates opportunities for optimization but also creates coupled failure modes. For example, an inaccurate AI model may select a poor route; a compromised telemetry source may manipulate that model; and an energy-saving policy may violate a latency guarantee.

6G: A Research Programme Rather Than a Single Product

6G is generally discussed as a future generation of mobile networking, but it is more accurately understood as a research programme involving radio technologies, distributed computing, sensing, automation, and new service models. Anticipated directions include sub-terahertz or terahertz experimentation, advanced multi-antenna systems, reconfigurable intelligent surfaces, integrated sensing and communication, semantic or goal-oriented communication, ubiquitous edge intelligence, and native integration of terrestrial and non-terrestrial access.

The attraction of higher frequency bands is increased available spectrum. The cost is severe propagation loss, blockage, limited range, and sensitivity to atmospheric absorption. Consequently, a high-frequency 6G system will require dense deployment, accurate beam management, intelligent control of surfaces or relays, and effective fallback to lower-frequency links. A claim of extremely high peak rate is therefore less significant than the distribution of user-experienced throughput under mobility, blockage, and load.

Integrated sensing and communication illustrates the broader 6G direction. Radio signals may simultaneously transport data and infer the position, movement, or shape of objects. This can support traffic management, industrial monitoring, and assisted navigation. However, sensing introduces privacy questions, false detections, adversarial manipulation, and difficult evaluation criteria. A system that improves localization but exposes sensitive movement patterns may not be acceptable in a regulated environment.

Research on semantic communication challenges the assumption that every bit must be delivered with equal importance. An application may require the meaning of a message—such as the position of an obstacle—rather than an exact reproduction of the original signal. Semantic methods can reduce traffic under constrained conditions, but they require application-specific models and a measurable definition of semantic fidelity. They also raise concerns about interoperability, explainability, and failures caused by a mismatch between sender and receiver knowledge.

A rigorous 6G experiment should therefore report more than peak data rate. Relevant measures include latency percentiles, jitter, reliability under blockage, energy per delivered bit or task, inference accuracy, coverage probability, control overhead, handover interruption, fairness among users, and resilience to failures. Results should be compared with strong 5G, Wi-Fi, or wired baselines rather than with an unrealistic zero-load reference.

Satellite and Non-Terrestrial Networks

Non-terrestrial networks (NTNs) use satellites, high-altitude platforms, or other airborne systems to extend or complement terrestrial connectivity. Low-Earth-orbit (LEO) systems offer shorter propagation delay than geostationary satellites and can provide broad coverage through constellations. Medium-Earth-orbit and geostationary systems offer different coverage, delay, and infrastructure trade-offs. High-altitude platforms may remain over a region for longer periods and can combine wide coverage with relatively flexible deployment.

The defining networking problem is mobility at several levels. A user terminal may move, a satellite may move rapidly relative to the ground, and the serving beam or gateway may change. Routing protocols must cope with predictable but frequent topology changes, intermittent feeder links, Doppler effects, handover decisions, and limited onboard processing. Delay is not only propagation delay; queueing, gateway selection, retransmission, and inter-satellite routing also matter.

NTNs are valuable for maritime communication, disaster recovery, rural access, aircraft connectivity, environmental monitoring, and backhaul. They are not automatically a replacement for terrestrial networks. Capacity per geographic area, terminal cost, weather sensitivity, regulatory restrictions, launch dependency, and gateway availability can constrain service quality. An integrated architecture should select the access technology according to application requirements rather than assuming that satellite access is always the most resilient option.

Experimental evaluation may combine orbital simulators, emulation of time-varying links, software-defined radios, and measurements from operational systems. A credible study should model visibility windows, link outages, Doppler shift, traffic demand, and gateway failures. It should also distinguish user-plane performance from control-plane convergence. A network that delivers packets successfully after a long route-recalculation interval may be unsuitable for interactive or safety-critical applications.

AI-Native and Autonomous Networking

AI is already used in networking for traffic prediction, anomaly detection, radio resource management, routing, energy control, and capacity planning. This is usually called AI-assisted networking: conventional protocols remain primary, while models support selected decisions. AI-native networking goes further by designing interfaces, data collection, control loops, and service abstractions around learning and inference from the beginning. Autonomous networking refers to the operational capability of sensing conditions, reasoning about policies, acting on the infrastructure, and verifying the result with limited human intervention.

A useful closed-loop model is:

observe → interpret → decide → act → verify → learn.

Telemetry provides observations; feature extraction and models interpret them; a policy selects an action; orchestration applies it; verification checks whether service objectives were preserved; and feedback updates the model or policy. The loop must include safety constraints. For example, a reinforcement-learning controller may optimize utilization but must not increase delay beyond a contractual bound.

AI-based control introduces risks that differ from ordinary software faults. Training data may be incomplete or biased toward normal operating conditions. Distribution shift may occur when traffic patterns, hardware, or topology changes. An adversary may poison data, evade an anomaly detector, or manipulate feedback. A model can also be correct on average while making unacceptable decisions for a minority of users. Explainability and rollback are particularly important when automated actions affect critical services.

A mature autonomous network uses graduated autonomy. Low-risk actions, such as adjusting a cache replica, may be automated; high-risk actions, such as changing security policy across a national backbone, may require approval. Digital twins can help by providing a controlled environment in which candidate actions are simulated before deployment. Yet a digital twin is only as useful as its model fidelity and update process. It cannot guarantee safety if important dependencies are absent from the model.

Evaluation should include not only prediction accuracy but also control quality, convergence time, constraint violations, recovery time, false-positive cost, fairness, energy use, and performance under adversarial or unfamiliar conditions. Reproducibility requires publishing datasets, feature definitions, baseline policies, random seeds where relevant, and the operational assumptions of the experiment.

Quantum Networking and Post-Quantum Security

Quantum networking seeks to distribute quantum states or entanglement among nodes. Potential applications include quantum key distribution, distributed quantum computing, and high-precision sensing. Quantum networks are not simply faster versions of classical networks. They depend on fragile quantum states, specialized hardware, synchronization, quantum memories, and repeaters. The no-cloning principle prevents direct copying of an unknown quantum state, while measurement can disturb the state. These properties make amplification and error recovery fundamentally different from classical packet forwarding.

A near-term quantum network is likely to coexist with classical control and data infrastructure. Classical channels are required for coordination, authentication, scheduling, and reporting measurement results. Quantum key distribution can provide a method for detecting certain forms of interception under specific assumptions, but it does not by itself solve endpoint compromise, denial of service, authentication, or poor implementation. Claims about “unbreakable” communication should therefore be treated critically.

Post-quantum cryptography (PQC) addresses a different problem: protecting classical cryptographic systems against adversaries that may eventually possess large-scale quantum computers. Quantum algorithms threaten widely used public-key mechanisms based on integer factorization or discrete logarithms. PQC replaces them with schemes based on mathematical problems believed to resist known quantum attacks, including lattice-, code-, hash-, and other constructions. These schemes may involve larger keys, signatures, or computational costs, so migration must consider bandwidth, latency, constrained devices, certificate infrastructure, and long-lived data.

Organizations should adopt crypto-agility: the capability to replace algorithms and parameters without redesigning every application. A migration programme begins with cryptographic inventory, identification of data requiring long-term confidentiality, assessment of protocol dependencies, testing of hybrid mechanisms, and staged deployment. The risk known as “harvest now, decrypt later” makes migration relevant even before a cryptographically capable quantum computer exists.

Information-Centric Networking

The conventional Internet primarily identifies endpoints. Information-centric networking (ICN) instead makes named content the central object. A request is directed toward a named piece of information, and the network may retrieve it from an origin server, cache, edge node, or other authorized source. This approach can improve distribution efficiency for popular and immutable content, especially in mobile, intermittent, or replicated environments.

ICN changes several architectural assumptions. Naming becomes a routing and security concern; security can be attached to the content through signatures rather than only to a communication channel; and caching becomes a first-class network function. The separation of content identity from location can support mobility and multi-source retrieval. However, name resolution, cache consistency, privacy, access control, cache pollution, and denial-of-service protection remain difficult. Signed content proves origin or integrity only when trust models and key management are correctly established; it does not automatically prove that content is current or authorized for every requester.

ICN is especially relevant to content distribution, vehicular networks, disaster communication, edge computing, and sensor data with well-defined names. Experimental studies should compare cache hit ratio, retrieval delay, bandwidth consumption, freshness, privacy leakage, and attack resistance against a well-configured content delivery or HTTP-based baseline. A high cache hit ratio is not sufficient if stale or sensitive content is served improperly.

Green Networking: Efficiency as a Systems Property

Green networking seeks to reduce energy consumption and environmental impact while maintaining acceptable service quality. Energy is used in radio access, switches, routers, optical systems, cooling, computation, manufacturing, and replacement. Consequently, optimizing only the active power of a router may shift rather than reduce environmental impact.

Common techniques include sleep modes, dynamic voltage and frequency scaling, energy-aware routing, workload consolidation, renewable-powered edge sites, efficient cooling, traffic shaping, and longer equipment lifetimes. Each has trade-offs. Putting equipment into sleep mode may increase wake-up delay and reduce redundancy. Consolidating workloads may save energy but create concentration risk. Routing through a renewable-powered site may increase latency or transit distance. A meaningful objective may therefore be multi-dimensional: minimize energy or carbon intensity subject to latency, reliability, capacity, and resilience constraints.

Researchers should report the system boundary and metric definition. Energy per bit, energy per useful task, total site energy, carbon emissions, and embodied carbon answer different questions. Measurements should include idle energy, cooling overhead, control traffic, and the effect of retransmissions. Green networking also has a rebound effect: cheaper or more efficient communication may stimulate additional traffic and offset some savings.

A Comparative Research Framework

The technologies discussed above can be compared through a common research workflow. First, define the problem and the service-level objective. “Improve the network” is not a testable question; “reduce 95th-percentile retrieval latency for replicated content under intermittent backhaul” is. Second, state the system boundary, assumptions, and baseline. Third, identify dependent variables such as latency, reliability, cost, energy, security loss, or fairness. Fourth, choose an evaluation method appropriate to the question.

Simulation provides scale and controlled comparison but may hide implementation overhead. Emulation executes real protocols in a controlled topology and is useful for repeatable experiments. Testbeds reveal hardware and radio effects but are expensive and limited in scale. Field trials provide ecological validity but are difficult to reproduce. Analytical models clarify relationships and bounds, although simplifying assumptions can become unrealistic. Strong research often combines these methods: analysis for insight, simulation for parameter exploration, emulation for protocol validation, and field measurements for external validity.

Experiments should include sensitivity analysis and failure cases. For AI controllers, evaluate distribution shift and adversarial inputs. For NTNs, vary visibility and gateway availability. For green systems, vary carbon intensity and traffic load. For PQC, measure handshake size and constrained-device cost. Statistical reporting should include distributions rather than only averages, confidence intervals where appropriate, and comparisons with meaningful baselines.

Illustrative Case Study: Resilient Connectivity for a Coastal Emergency Network

Background

Consider a fictional coastal region exposed to cyclones. Its normal communication system combines terrestrial 5G, fibre backhaul, edge computing, and public safety services. Flooding can disable power, fibre routes, and local access sites. Emergency managers propose an integrated network using satellite access, autonomous traffic engineering, edge caching, and post-quantum protection for sensitive records.

Situation and Problem

During a major storm, the region must support emergency voice and telemetry, public information, hospital coordination, and sensor data. The network must operate with reduced terrestrial capacity. A design that maximizes aggregate throughput may still fail if emergency traffic is delayed or if automated decisions consume scarce satellite bandwidth.

Analysis

The service classes are separated by requirements. Emergency control traffic requires high reliability and bounded delay; public information is cacheable; sensor uploads can tolerate delay; confidential records require authenticated and quantum-resistant cryptographic protection. Satellite links provide coverage but have variable capacity and handover conditions. Edge caches can reduce repeated downloads of public instructions. An autonomous controller can redirect traffic, but it must operate within explicit policies and retain a human override.

Approach

The proposed architecture uses terrestrial links whenever available and satellite links as a resilient complement. Named public advisories are replicated at edge and satellite-connected caches. A policy-based controller monitors link status and assigns priority rather than allowing an unconstrained learning model to decide. The controller is tested in an emulated storm topology with gateway failures, changing demand, and intermittent power. PQC is introduced first for management channels and long-lived sensitive data, with hybrid cryptographic operation during migration.

Evaluation metrics include emergency-message delivery probability, 95th-percentile delay, cache hit ratio, recovery time after gateway failure, energy consumed per delivered service unit, cryptographic handshake overhead, and the number of policy violations. A field pilot is then conducted with non-critical traffic and synthetic emergency workloads.

Outcome

The fictional evaluation shows that satellite integration improves coverage and recovery, but cannot alone provide terrestrial-level capacity during simultaneous regional demand. Caching substantially reduces repeated backhaul traffic, while strict priority policies protect emergency traffic. Autonomous control improves recovery time only when telemetry remains trustworthy and fallback rules are available. PQC increases control-message size, making fragmentation and certificate management important engineering concerns.

Learning Points

The case demonstrates that resilience is architectural rather than synonymous with adding one technology. Satellite, caching, AI control, and PQC solve different problems. Their interaction must be evaluated under failure, overload, security attack, and energy constraints. It also shows why service-specific metrics are more informative than a single network-wide throughput value.

Challenges, Governance, and Professional Responsibilities

Emerging networks create governance issues concerning spectrum, orbital resources, cross-border data, lawful access, safety, environmental impact, and accountability for automated decisions. Satellite constellations may create debris and astronomical interference concerns. Sensing-enabled networks may collect information about people without clear consent. AI systems may reproduce unfair treatment through prioritization or automated denial of service. Quantum and post-quantum technologies can affect national security and long-term confidentiality.

Common professional mistakes include presenting laboratory peak performance as deployable capability, ignoring control-plane overhead, comparing technologies with unequal assumptions, treating AI accuracy as operational success, and reporting energy savings without defining the system boundary. Another error is designing security as a later add-on. In emerging architectures, naming, telemetry, orchestration, hardware identity, and cryptographic agility should be considered from the beginning.

Best practice is to combine technical evaluation with threat modelling, lifecycle analysis, reproducible experimentation, independent validation, and explicit uncertainty. Engineers should document assumptions, disclose negative results, preserve audit logs for autonomous actions, and provide safe degradation and rollback. Organizational adoption should proceed through pilots with measurable exit criteria rather than through technology-driven procurement.

Current Research Directions

Important research directions include integrated terrestrial–satellite orchestration; energy- and carbon-aware network control; trustworthy machine learning with uncertainty estimates; federated and privacy-preserving network intelligence; semantic metrics that are application-independent; quantum repeater engineering; practical PQC migration; content-centric mobility and access control; and digital twins validated against operational data. Cross-layer research is especially significant because radio, transport, computing, security, and energy decisions increasingly influence one another.

The most valuable contributions will not necessarily be systems with the highest theoretical performance. They will be architectures and methods that are measurable, secure, interoperable, energy-conscious, and robust when assumptions fail. Emerging networking research is therefore as much about disciplined evaluation and responsible deployment as it is about novel protocols.

Chapter Summary

Emerging networks extend beyond faster packet forwarding. 6G investigates highly adaptive radio, sensing, semantic communication, edge intelligence, and integrated access. Non-terrestrial networks provide coverage and resilience but introduce mobility, delay, capacity, and regulatory challenges. AI-native and autonomous networking use closed-loop observation and control, requiring safety constraints, trustworthy telemetry, and evaluation under unfamiliar conditions. Quantum networking has specialized physical requirements, while post-quantum cryptography addresses the future risk to classical public-key systems and demands crypto-agile migration. Information-centric networking changes the Internet’s focus from endpoints to named content, and green networking treats energy and carbon as system-level objectives. Across all areas, credible research depends on explicit problem definitions, meaningful baselines, appropriate experimental methods, security and governance analysis, and metrics that reflect real service quality.

Key Terms

6G: A research direction for future mobile systems integrating advanced radio, sensing, intelligence, and heterogeneous connectivity.

Non-terrestrial network: A network using satellites or airborne platforms to provide or complement communication services.

LEO constellation: A group of low-Earth-orbit satellites designed to provide broad or continuous coverage.

Integrated sensing and communication: Joint use of communication signals for data transfer and environmental or positional sensing.

Semantic communication: Communication designed to convey task-relevant meaning rather than reproduce every source symbol.

AI-assisted networking: Use of AI to support selected networking decisions within predominantly conventional architectures.

AI-native networking: Networking designed from the outset around learning, inference, and data-driven control.

Autonomous network: A network capable of sensing, deciding, acting, and verifying with limited human intervention.

Digital twin: A computational representation of a physical network used for monitoring, analysis, or controlled experimentation.

Quantum networking: Networking that distributes quantum states or entanglement among quantum-capable nodes.

Post-quantum cryptography: Cryptography designed to resist attacks from known classical and quantum algorithms.

Crypto-agility: The ability to replace cryptographic algorithms and parameters with limited system redesign.

Information-centric networking: An architecture that identifies and retrieves named information rather than primarily addressing endpoints.

Content-centric security: Protection attached to content through mechanisms such as signatures and integrity verification.

Green networking: Design and operation of networks to reduce energy use and environmental impact while meeting service objectives.

Energy per bit: Energy consumed to transmit or process a quantity of data, subject to a defined system boundary.

Distribution shift: A change between the data conditions used to develop a model and those encountered during operation.

Service-level objective: A measurable target for characteristics such as latency, reliability, availability, or throughput.

Crypto migration: The planned transition from existing cryptographic mechanisms to replacement mechanisms.

Research reproducibility: The ability of others to repeat an experiment using documented methods, data, and assumptions.

Review Questions

Short Answer Questions

1. Why is 6G better understood as a research programme than as a single technology?

2. Distinguish AI-assisted networking from AI-native networking.

3. What mobility problems are characteristic of non-terrestrial networks?

4. Why does post-quantum cryptography require crypto-agility?

5. State two limitations of information-centric networking.

Long Answer Questions

1. Explain the architecture, opportunities, and limitations of integrated terrestrial–non-terrestrial networks.

2. Describe the observe–interpret–decide–act–verify–learn loop and discuss safeguards for autonomous networking.

3. Compare quantum networking with post-quantum cryptography in terms of purpose, assumptions, and deployment challenges.

4. Discuss how information-centric networking changes naming, caching, security, and mobility.

5. Develop a research methodology for evaluating the energy and service-quality trade-offs of a green network.

Analytical / Application Questions

1. A rural emergency network has intermittent fibre, satellite backup, and limited battery capacity. Design a service-aware architecture and specify the metrics and failure scenarios for its evaluation.

2. An AI routing controller improves average throughput but increases 99th-percentile delay for remote users. Analyse whether deployment is justified and propose additional tests or constraints.

3. An organization operates long-lived confidential records and constrained IoT devices. Construct a phased post-quantum migration plan and identify the principal performance and governance risks.

22. Multiple Choice Questions with Answers

Question 1

Which evaluation measure is most appropriate for assessing a 6G system operating under blockage and mobility?

A. Peak theoretical data rate only

B. Average packet size only

C. Latency percentiles, reliability, energy, and handover interruption

D. Number of protocol headers only

Correct Answer: C. Latency percentiles, reliability, energy, and handover interruption

Question 2

What is a principal networking challenge introduced by LEO satellite constellations?

A. A completely static topology

B. Frequent topology and serving-beam changes

C. The elimination of propagation delay

D. The impossibility of using gateways

Correct Answer: B. Frequent topology and serving-beam changes

Question 3

In an autonomous network, the verification stage primarily determines whether:

A. The selected action achieved objectives without violating constraints

B. All packets have identical payloads

C. A satellite has become geostationary

D. A cryptographic key has been copied

Correct Answer: A. The selected action achieved objectives without violating constraints

Question 4

Which statement correctly distinguishes post-quantum cryptography from quantum networking?

A. PQC distributes entanglement, while quantum networking replaces certificates

B. PQC protects classical systems against quantum-capable attacks, while quantum networking uses quantum states between nodes

C. Both terms describe satellite routing protocols

D. Quantum networking is only an energy-efficiency technique

Correct Answer: B. PQC protects classical systems against quantum-capable attacks, while quantum networking uses quantum states between nodes

Question 5

Why is crypto-agility important in post-quantum migration?

A. It permits algorithm replacement without redesigning every application

B. It guarantees that endpoints cannot be compromised

C. It removes the need for authentication

D. It reduces all public-key messages to one byte

Correct Answer: A. It permits algorithm replacement without redesigning every application

Question 6

Which feature is most characteristic of information-centric networking?

A. Content is requested by name and may be obtained from an authorized cache

B. Every object must be retrieved from one permanent host

C. Routing is based only on physical cable length

D. Content signatures automatically solve freshness and authorization

Correct Answer: A. Content is requested by name and may be obtained from an authorized cache

Question 7

A green-networking study reports lower router power but excludes cooling, retransmissions, and idle equipment. What is the main methodological problem?

A. The study uses too many service metrics

B. The system boundary is incomplete

C. The network must be quantum-enabled

D. Peak throughput is always irrelevant

Correct Answer: B. The system boundary is incomplete

Question 8

Which experimental method offers real protocol execution in a controlled, repeatable topology without requiring a full-scale field deployment?

A. Emulation

B. Pure theoretical analysis

C. Uncontrolled observation

D. Cryptographic inventory

Correct Answer: A. Emulation

Question 9

In the emergency-network case, why are public advisories suitable for edge or satellite caching?

A. They are often repeated and can be replicated to reduce backhaul demand

B. They must always remain private to one endpoint

C. Caching eliminates all satellite handovers

D. Caching guarantees unlimited capacity

Correct Answer: A. They are often repeated and can be replicated to reduce backhaul demand

Question 10

An AI network controller performs well in training but fails after traffic patterns change. This is an example of:

A. Content naming

B. Distribution shift

C. Quantum entanglement

D. Dynamic voltage scaling

Correct Answer: B. Distribution shift