Adopting artificial intelligence (AI) in radio access networks (RANs) presents several challenges, limited availability of link-level measurements (e.g., CQI reports), stringent real-time processing constraints (e.g., sub-1ms per TTI), and network heterogeneity (different spectrum bands, cell types, and vendor equipment). A critical yet often overlooked barrier lies in the computational and memory limitations of RAN baseband hardware—particularly in legacy 4thGeneration (4G) systems—which typically lack on-chip neural accelerators. As a result, only lightweight AI models (under 1Mb and sub-100µs inference time) can be effectively deployed, limiting both their performance and applicability. However, achieving strong generalization across diverse network conditions often requires large-scale models with substantial resource demands. To address this trade-off, this paper investigates policy distillation in the context of a reinforcement learning–based link adaptation task. We explore two strategies: single-policy distillation, where a scenario-agnostic teacher model is compressed into one generalized student model; and multi-policy distillation, where multiple scenario-specific teachers are consolidated into a single generalist student. Experimental evaluations in a high-fidelity, 5thGeneration (5G)-compliant simulator demonstrate that both strategies produce compact student models that preserve the teachers’ generalization capabilities while complying with the computational and memory limitations of existing RAN hardware.

While the problems of sum-rate maximization and sum-power minimization subject to quality of service (QoS) constraints in the multiple input multiple output interference broadcast channel (MIMO IBC) have been widely studied, most of the proposed solutions have neglected the user scheduling aspect assuming that a feasible set of users has been previously selected. However, ensuring QoS for each user in the MIMO IBC involves the joint optimization of transmit/receive beamforming vectors, transmit powers, and user scheduling variables. To address the full problem, we propose a novel formulation of a rate-constrained sum-utility maximization problem which allows to either deactivate users or minimize the QoS degradation for some scheduled users in infeasible scenarios. Remarkably, this is achieved avoiding the complexity of traditional combinatorial formulations, but rather by introducing a novel expression of the QoS constraints that allows to solve the problem in a continuous domain. We propose centralized and decentralized solutions, where the decentralized solutions focus on practical design and low signaling overhead. The proposed solutions are then compared with benchmarking algorithms, where we show the effectiveness of the joint scheduling and transceiver design as well as the flexibility of the proposed solution performing advantageously in several MIMO IBC scenarios.

Optimizing radio transmission power and user data rates in wireless systems requires full system observability. While the problem has been extensively studied in the literature, practical solutions approaching optimality exploiting only the partial observability available in real systems are still lacking. This paper proposes a reinforcement learning approach to downlink power control and rate adaptation in cellular networks that closes this gap. We present a comprehensive design of the learning framework that includes the characterization of the system state, a general reward function, and an efficient learning algorithm. System level simulations show that our design quickly learns a power control policy that brings significant energy savings and fairness across users in the system.

Superposed transmission on the broadcast channel delivers multiple private messages simultaneously to the intended users. Superposed transmission is employed in non-orthogonal multiple access to increase the aggregate throughput and connectivity of wireless cellular systems. Rate-adaptive constellation Expansion Multiple Access (REMA, a.k.a. RA-CEMA) is a non-orthogonal multiple access scheme which exhibits performance similar to the conventional superposed transmission schemes for transmission on the degraded broadcast channel, while requiring less complex signal processing and scheduling algorithms. This paper solves the problem of user scheduling and rate allocation for REMA. We pose the problem as a weighted sum-rate optimization and develop provably optimal algorithms for user selection and rate allocation with REMA broadcast-channel transmission. System-level evaluations show that our schemes outperform an LTE-A compliant orthogonal multiple access system and exhibit similar or better performance compared to other known non-orthogonal multiple access schemes.

To meet the increasing data rate demand, radio networks are becoming increasingly heterogeneous as we move towards 5G. In step with this, we present new algorithms for cellular-controlled LTE/WLAN traffic steering and link aggregation using the newly standardized WLAN support within the 3GPP LTE-A Rel-13. The proposed schemes rely on user measurement reports for estimating the WLAN data rate before steering users to WLAN. We evaluate their performance in several scenarios against baseline solutions compliant with Rel- 12 specifications for cellular-assisted WLAN offloading. We show that a proper user selection for WLAN offloading is crucial to fully harvest the benefits of cellular-controlled LTE/WLAN interworking, and that outdoor WLAN and indoor co-located LTE/WLAN deployments are the most promising for enjoying the Rel-13 enhancements.

In LTE Release 12, Dual Connectivity promises flexible delivery of downlink data to the mobile device from macro and small cells, connected by non-ideal backhaul. This paper considers how user experience can benefit from easier load balancing between a macro cell and multiple associated small cells (because L3 handover is not required) and from bearer splitting, where data can exploit both layers simultaneously. Algorithms are proposed and tested with video traffic in Outdoor and Indoor & Outdoor scenarios, using a system simulator. Load balancing and/or bearer splitting algorithms that target weakly performing users, when applied on top of a baseline SINR cell association, give similar performance, with a marked improvement in the user satisfaction rate.

The next generation of radio access networks is expected to offer users flexible radio access to a wide range of frequency bands and radio access technologies (RATs) in dense deployments. In this context, we investigate novel lightweight algorithms for spectrum access in ultra-dense networks. We pose the problem as a network utility maximization and apply Lagrange duality to devise distributed algorithms for users to jointly select access nodes and spectrum bands. Leveraging our theoretical framework, we further propose a heuristic scheme for spectrum access based on the expected long-term user data rate per RAT, hence on a measure of the RATs' traffic load. Numerical examples show that our schemes achieve significant throughput gains on top of the gains achievable by network densification.

Densification of the radio access network and massive connectivity are envisaged to significantly increase the capacity of future radio access networks, well beyond 4G capabilities. This comes at the price of a harsher interference environment and an increased complexity of radio resource management (RRM), to the extent that today's RRM protocols may no longer be efficient. In this article we propose an inter-cell interference coordination mechanism that departs from the conventional concept of trading cell-centre spectral efficiency for cell-edge spectral efficiency, but rather aims at improving spectral efficiency in the entire service area. This is achieved by coordinating and optimizing access point transmission blanking on a fast time-scale while scheduling mobile stations in the direction of least interference. Simulation results indicate that the spectral efficiency can effectively be improved over the entire service area, with up to 200% increase of cell-edge users capacity and an increased average network capacity up to 30% compared to traditional reuse-1 schemes.

Opportunistic routing is widely known to have substantially better performance than unicast routing in wireless networks with lossy links. However, wireless sensor networks are usually duty cycled, that is, they frequently enter sleep states to ensure long network lifetime. This renders existing opportunistic routing schemes impractical, as they assume that nodes are always awake and can overhear other transmissions. In this article we introduce ORW, a practical opportunistic routing scheme for wireless sensor networks. ORW uses a novel opportunistic routing metric, EDC, that reflects the expected number of duty-cycled wakeups that are required to successfully deliver a packet from source to destination. We devise distributed algorithms that find the EDC-optimal forwarding and demonstrate using analytical performance models and simulations that EDC-based opportunistic routing results in significantly reduced delay and improved energy efficiency compared to traditional unicast routing. In addition, we evaluate the performance of ORW in both simulations and testbed-based experiments. Our results show that ORW reduces radio duty cycles on average by 50% (up to 90% on individual nodes) and delays by 30% to 90% when compared to the state-of-the-art.

Convergecast, in which data from a set of source devices is delivered to a single data sink, is a critical functionality in networks deployed for industrial monitoring and control. We address the latency-optimal link scheduling problem for convergecast in networks operating according to the recent WirelessHART standard. When there is no restriction on the number of channels, we present a latency-optimal scheduling policy in which each routing node is required to buffer at most one packet at any point in time. For networks with a limited number of channels, we first establish a lower bound on the number of channels for latency-optimal convergecast and a lower bound on latency for convergecast using a fixed number of channels, and then present a heuristic scheme for channel-constrained latency-optimal convergecast scheduling. Simulation results confirm that, at much modest computational cost, our heuristic scheme can construct convergecast schedules with latency close to that of the optimal schedules.

Multichannel communication has been proposed as alternative to adaptive (single-channel) routing protocols for mitigating the impact of interference and link dynamics in wireless sensor networks. While several studies have advocated features of both techniques (not without running up against contradicting arguments) a comprehensive study that aligns these results is still lacking. This paper aims at filling this gap. We present an experimental test bed setup used to perform extensive measurements for both single-channel and multichannel communication. We first analyze single-channel and multichannel communication over a single-hop in terms of packet reception ratio, maximum burst loss, temporal correlation of losses, and loss correlations across channels. Results show that multichannel communication with channel hopping significantly reduces link burstiness and packet loss correlation. For multi-hop networks, multi-channel communication and adaptive routing show similar end-to-end reliability in dense topologies, while multichannel communication can outperform adaptive routing in sparse networks with bursty links.

Interference and link dynamics constitute great concerns for stability and performance of protocols in WSNs. In this paper we evaluate the impact of channel hopping and adaptive routing on delay and reliability focusing on delay critical applications.

Opportunistic routing is widely known to have substantially better performance than traditional unicast routing in wireless networks with lossy links. However, wireless sensor networks are heavily duty-cycled, i.e. they frequently enter deep sleep states to ensure long network life-time. This renders existing opportunistic routing schemes impractical, as they assume that nodes are always awake and can overhear other transmissions. In this paper, we introduce a novel opportunistic routing metric that takes duty cycling into account. By analytical performance modeling and simulations, we show that our routing scheme results in significantly reduced delay and improved energy efficiency compared to traditional unicast routing. The method is based on a new metric, EDC, that reflects the expected number of duty cycled wakeups that are required to successfully deliver a packet from source to destination. We devise distributed algorithms that find the EDC-optimal forwarding, i.e. the optimal subset of neighbors that each node should permit to forward its packets. We compare the performance of the new routing with ETX-optimal single path routing in both simulations and testbed-based experiments.

This paper studies the problem of optimal forwarding for reliable and energy-efficient real-time communication over multi-hop wireless lossy networks. We impose a strict per-packet latency bound and develop forwarding policies that maximize the probability that the packet is delivered within the specified deadline minus a transmission energy cost. A solution to this problem allows to characterize the set of achievable latency-reliability pairs and to trace out the Pareto frontier between achievable deadline-constrained reliability and transmission energy cost. We develop dynamic programming-based solutions under a finite-state Markov channel model. Particular instances with Bernoulli and Gilbert-Elliot loss models that admit numerically efficient solutions are discussed and our results are demonstrated on several examples.

Achieving low-power operation in wireless sensor networks with high data load or bursty traffic is challenging. The hidden terminal problem is aggravated with increased amounts of data in which traditional backoff-based contention resolution mechanisms fail or induce high latency and energy costs. We analyze and optimize Strawman, a receiver-initiated contention resolution mechanism that copes with hidden terminals. We propose new techniques to boost the performance of Strawman while keeping the resolution overhead small. We finally validate our improved mechanism via experiments.

The traditionally wired automation infrastructure is quickly migrating to more flexible and scalable wireless solutions. To cope with the stringent requirements of process automation in terms of latency and reliability, the network resources must be optimized to ensure timely and reliable communication. This paper considers the joint routing and transmission scheduling problem for reliable real-time communication over lossy networks. Specifically, we impose a strict latency bound for packet delivery from source to destination, and devise optimal transmission scheduling policies that maximize the success probability of delivering the packet within the specified deadline. A solution to this problem allows to characterize the set of achievable latencies and packet reliability for a given network. We offer a complete understanding of the problem when erasure events on links are independent and follow a Bernoulli process. We consider both static and dynamic resource allocation policies, and compare them in numerical examples.

The increased industrial interest in wireless sensor networks demands a shift from optimizing protocols for energy-efficient reporting of sporadic events to developing solutions for high-rate real-time data collection and dissemination. We study time-optimal convergecast under the communication constraints of commodity sensor network platforms, and propose a novel convergecast model in which packet copying between the microcontroller and the radio transceiver is separated from packet transmission, thereby improving channel utilization and system throughput. Based on this model, we establish tight lower bound on the number of time slots for convergecast in networks with tree routing topology, and present both centralized and distributed algorithms for generating time-optimal convergecast schedules. Our scheme is also memory-efficient as each node needs to buffer at most one packet at any time. We evaluate our scheme in simulation and on real hardware, and show that our scheme can achieve a throughput of 203 kbit/s (86.4% of the theoretical upper bound) and up to 86.24% improvement compared with traditional TDMA-based convergecast. With optimal routing tree and maximum MAC layer payload, convergecast in a network with 20 sensor nodes can be completed in only 100 ms.

Network multiple input, multiple output (MIMO) systems are built around a broadband backbone network that allows for the fast communication of channel state information (CSI) as well as user data between different base stations. Previous works have shown that multicell channel adaptive (opportunistic) power control can minimize the sum power or maximize the sum rate when the backbone is used for the exchange of CSI in network MIMO systems. In this work we investigate the gains of multicell opportunistic power control under per user fairness constraints when both CSI and user data are shared between multiple sites. We find that multicell opportunistic power control working in concert with uplink joint signal detection is an efficient means both for the capacity and the power control problems that not only minimizes sum power or maximizes overall capacity, but is also able to provide arbitrary level of fairness.

This paper describes a methodology for controller and communication scheduling co-design in control systems operating over wirelessHART networks. Data collection and dissemination operations are identified and scheduled to minimize the nominal communication latency. Techniques for improving the reliability of the network when link transmissions are unreliable are discussed, and a Markov-chain model for computing the latency distribution of data collection operations for a given schedule is proposed. The resulting latency models allow to represent the networked control loop as a jump-linear system, whose performance can be analyzed using techniques from stochastic control. We demonstrate how this framework can be used to co-design a networked LQG controller for a five-by-five MIMO control loop.

We consider the problem of setting the uplink signal-to-noise-and-interference (SINR) target and allocating transmit powers for mobile stations in multicell spatial multiplexing wireless systems. Our aim is twofold: to evaluate the potential of such mechanisms in coordinated multipoint transmission (CoMP) systems, and to develop scalable numerical schemes that allow real-time near-optimal resource allocation across multiple sites. We formulate two versions of the SINR target and power allocation problem: one for maximizing the sum rate subject to power constraints, and one for minimizing the total power needed to meet a sum-rate target. To evaluate the potential of our approach, we perform a semi-analytical study in Mathematica using the augmented Lagrangian penalty function method. We find that the gain of the joint optimum SINR setting and power allocation may be significant depending on the degree of fairness that we impose. We develop a numerical technique, based on successive convexification, for real-time optimization of SINR targets and transmit powers. We benchmark our procedure against the globally optimal solution, and demonstrate consistently strong performance in realistic CoMP scenarios.

Current proposals for joint power and rate allocation protocols in ad hoc networks require a large signaling overhead, and do not adhere to the natural time-scales of transport and power control mechanisms. We present a solution that overcomes these issues. We pose the protocol design as a network utility maximization problem and adopt primal decomposition techniques to devise a novel distributed cross-layer design for transport and physical layer that achieves the optimal network operation. Our solution has several attractive features compared to alternatives: it adheres to the natural time-scale separation between rapid power control updates and slower end-to-end rate adjustments; it allows simplified power control mechanisms with reduced signalling requirements, and distributed slow rate cross-layer signalling mechanisms; and it maintains feasibility at each iteration. We validate the theoretical framework and compare the solution alternatives with numerical examples.

This paper presents a novel cross-layer design for joint power and end-to-end rate control optimization in DS-CDMA wireless networks, along with a detailed implementation and evaluation in the network simulator ns-2. Starting with a network utility maximization formulation of the problem, we derive distributed power control, transport rate and queue management schemes that jointly achieve the optimal network operation. Our solution has several attractive features compared to alternatives: it adheres to the natural time-scale separation between rapid power control updates and slower end-to-end rate adjustments, and uses simplified power control mechanisms with reduced signalling requirements. We argue that these features are critical for a successful real-world implementation. To validate these claims, we present a detailed implementation of a cross-layer adapted networking stack for DS-CSMA ad-hoc networks in ns-2. We describe several critical issues that arise in the implementation, but are typically neglected in the theoretical protocol design, and evaluate the alternatives in extensive simulations.

A generalized analysis of the IEEE 802.15.4 medium access control (MAC) protocol in terms of reliability, delay and energy consumption is presented. The IEEE 802.15.4 exponential backoff process is modeled through a Markov chain taking into account retry limits, acknowledgements, and unsaturated traffic. Simple and effective approximations of the reliability, delay and energy consumption under low traffic regime are proposed. It is demonstrated that the delay distribution of IEEE 802.15.4 depends mainly on MAC parameters and collision probability. In addition, the impact of MAC parameters on the performance metrics is analyzed. The analysis is more general and gives more accurate results than existing methods in the literature. Monte Carlo simulations confirm that the proposed approximations offer a satisfactory accuracy.

The introduction of relaying techniques into cellular networks is expected to reduce the total infrastructure cost, especially when coverage extension is sought. With the advent of networks such as LTE, guaranteeing high data-rate coverage may become a challenge even for incumbent operators which nowadays provide full coverage for voice service. However it is not straightforward that operators will be able (or want) to guarantee coverage for the high data-rates. In this paper we study how the techno-economic viability of the relaying solution depends on the type of service to be provided by the operator. We exemplify the trade off between coverage and system throughput with two fairness criteria: perfect fairness (coverage guarantee) and proportional fairness. We show that relays provide advantages when the operator is interested in providing bit-rate/QoS guarantees. When maximizing the system throughput or cell capacity, relays are of less value.

Convergecast, in which data from a set of sources is routed toward one data sink, is a critical functionality for wireless networks deployed for industrial monitoring and control. We address the joint link scheduling and channel assignment problem for convergecast in networks operating according to the recent WirelessHART standard. For a linear network with N single-buffer devices, we demonstrate that the minimum time to complete convergecast is 2N-1 time-slots, and that the minimum number of channels required for this operation is lceilN/2rceil. When the devices are allowed to buffer multiple packets, we prove that the optimal convergecast time remains the same while the number of required channels can be reduced to . For both cases, we present jointly time- and channel-optimal scheduling policies with complexity O(N2). Numerical results demonstrate that our schemes are also efficient in terms of memory utilization.

This paper considers distributed protocol design for joint sub-carrier, transmission scheduling and power management in uplink/downlink multi-cellular OFDMA wireless networks. The optimal solution to this problem is hard to achieve, both in theory and in practice. We propose a fully decentralized resource allocation scheme combining decomposition methods for convex optimization with a strategic non-cooperative game formulation of the power and sub-carrier allocation subproblem. Although the final protocols are suboptimal, they drastically reduce computation time while maintaining the overall system performance close to the optimal, exhibiting strong robustness to multiple access interference. We validate the theoretical framework and quantify the performance with numerical examples.

We present an iterative joint scheduling-routing algorithm for characterizing the long-term performance of a cellular-relaying network. The physical layer model is based on ideal rate adaptation, fixed transmission power, and average interference. At the MAC layer, time-shares of a common channel are allocated to links in a CSMA/CA-like fashion. At the transport layer, one or more parallel routes can transfer the data flow from a source to the destination, and the average end-user rates are adjusted so as to maximize a global utility function. The general problem of allocating time-shares and selecting link rates is nonconvex. We propose an iterative algorithm and a novel approximation for the physical layer so as to obtain a convex formulation at each iteration step. If only one route is available per source, the problem admits a convex equivalent formulation and the iterative algorithm converges to the global optimum. If more routes per source are available, the problem remains nonconvex. However we show in a simple tractable example that the joint formulation yields sum-log-utility values comparable to the results obtained through exhaustive search of all possible route combinations.

The enhancement of cellular networks with relaying technologies is expected to bring significant techno-economic benefits, but proper allocation of resources may be a challenging task. This paper addresses the question whether cross-layer optimization is essential for obtaining the expected relaying benefits. We present a network dimensioning approach based on joint routing and radio resource allocation. We consider a macro-cellular scenario with a max-min user-rate objective. If a single route per source is fixed, rates and time-shares can be optimally allocated to each link. In a simple macro-cellular propagation environment, the relays must cost at most 10-25% of a base station cost, if the network is deployed at once. The results are only slightly better when the routing is included in the joint resource allocation.

We rate optimization and radio resource management in wireless OFDMA-based mesh networks. Radio units equipped with multiple radio interfaces are combined with the OFDMA medium access scheme to make up for the classical limitations of single- carrier wireless networks. We pose the problem as a utility maximization problem subject to link-rate constraints, power control and transmission scheduling in terms of time slots, channels and radio interfaces. The optimal solution to this problem is in general hard to achieve, both in theory and in practice. We propose an alternative solution approach that drastically reduces the computation time, and suggest a heuristic scheme that attempts to reduce the computation time even further while maintaining the overall system performance close to the optimal. For each scheme we validate the theoretical framework and quantify the performance with numerical examples.

We consider the problem of joint end-to-end bandwidth allocation and radio resource management in WiMax single-carrier wireless networks. The design problem is posed as a utility maximization problem subject to link rate constraints which involve transmission scheduling and power allocation. Inspired by a centralized algorithm for solving the associated optimization problem, we proceed systematically in our development of distributed resource allocation mechanisms. Contrary to the centralized algorithm, the proposed solution is distributed and of low computational complexity, generates schedules of finite length and with fixed time-slot durations, and acts on local neighborhood information only. Although the final scheme is suboptimal, we isolate and quantify the performance losses incurred and demonstrate strong performance in examples.

We consider the problem of designing distributed mechanisms for joint congestion control and resource allocation in spatial-reuse TDMA wireless networks. The design problem is posed as a utility maximization subject to link rate constraints that involve both power allocation and transmission scheduling over multiple time-slots. Starting from the performance limits of a centralized optimization based on global network information,we proceed systematically in the development of distributed and transparent protocols. In the process,we introduce a novel decomposition method for convex optimization,establish its convergence for the utility maximization problem and demonstrate how it suggests a distributed solution based on flow control optimization and incremental updates of the transmission schedule.We develop a two-step procedure for finding the schedule updates and suggest two schemes for distributed channel reservation and power control under realistic interference models. Although the final protocols are suboptimal,we isolate and quantify the performance losses incurred by each simplification and demonstrate strong performance in examples.

Network performance can be increased if the traditionally separated network layers are jointly optimized. Recently, network utility maximization has emerged as a powerful framework for studying such cross-layer issues. In this paper, we review and explain three distinct techniques that can be used to engineer utility-maximizing protocols: primal, dual, and cross decomposition. The techniques suggest layered, but loosely coupled, network architectures and protocols where different resource allocation updates should be run at different time-scales. The decomposition methods are applied to the design of fully distributed protocols for two wireless network technologies: networks with orthogonal channels and network-wide resource constraints, as well as wireless networks where the physical layer uses spatial-reuse time-division multiple access. Numerical examples are included to demonstrate the power of the approach

We present an efficient approach for cross-layer optimization of a class of wireless networks equipped with multi-user detectors. For a given network, the method provides the optimal operation of transport, routing and radio link layers as well as the optimal coordination across the layers. The algorithm combines a column generation procedure with an optimal greedy resource allocation scheme into a powerful numerical method for computing the performance limits for networks of significant sizes. The method is applied to a simple network scenario and performance benefits of multi-user detectors and cross-layer coordination over alternative schemes are investigated.