The increasing demand for energy-aware, yet service-assured networking has expanded the importance of Software-Defined Networks (SDN), where it emerged as a promising paradigm to improve the flexibility of communication networks by separating the network control plane from the data plane. However, achieving energy-efficient routing while guaranteeing Quality of Service (QoS) remains a challenging task for effective network management, as traditional algorithms fail to adapt to dynamic traffic demands. This paper introduces the Double critics with Priority Replay Buffer DDPG agent-based Routing framework (DP-DRRouting), an energy-efficient and QoS-aware routing architecture for SDN. Unlike existing Reinforcement Learning (RL)-based solutions that suffer from scalability limitations, unstable learning behavior, and slow convergence, particularly in dynamic environments. The proposed architecture core is a Double-Critic with Priority Replay Buffer Deep Deterministic Policy Gradient (DP-DDPG) agent that reduces reward overestimation and enhances stability. To further enhance adaptability, we integrate a direct-policy transfer learning mechanism that enables the agent to fine-tune its knowledge under topology changes without retraining from scratch. DP-DRRouting operates on path-level decisions, optimizing simultaneously for delay, packet loss ratio, throughput, and energy efficiency. The results show that DP-DRRouting achieves up to 12% energy efficiency and 50% throughput improvement compared to Open Shortest Path First (OSPF), while adapting to link failures with 40% throughput gains. These improvements confirm that the proposed framework provides a stable and adaptive routing solution that achieves near-optimal performance under varying traffic loads and topology changes.

Abstract The reliance on wireless services to exchange critical data is associated with various threats and attacks, which must be mitigated to ensure integrity and security of those wireless services. Posing a serious challenge to wireless systems, jamming is among these attacks. In order to mitigate jamming, directive antennas are used to minimize the signals that are received from the jammer, while maximizing the received legitimate signal from the authorized transmitter. In this paper, we propose a machine learning‐based anti‐jamming framework to provide a spatially dynamic and instantaneous anti‐jamming performance that is achieved at the physical layer. The proposed framework incorporates a dataset that can be deployed in the hardware of a receiver with a massive Multiple‐Inputs Multiple‐Outputs–(MIMO) antenna. Our extensive performance evaluation results demonstrate the effective performance of the proposed framework in preserving integrity of a massive–MIMO communication system despite the presence of a hostile jammer. Particularly, due to the tabular nature of the generated dataset, tree‐based random forest models achieved the best performance with a signal‐to‐interference‐plus‐noise ratio accuracy of and fast anti‐jamming response in around 1.66 s under sever jamming conditions.

Utilizing the intelligence at the network edge, edge computing paradigm emerges to provide time-sensitive computing services for Internet of Things. In this paper, we investigate sustainable computation offloading in an edge-computing system that consists of energy harvesting-enabled mobile devices (MDs) and a dispatcher. The dispatcher collects computation tasks generated by IoT devices with limited computation power, and offloads them to resourceful MDs in exchange for rewards. We propose an online Rewards-optimal Auction (RoA) to optimize the long-term sum-of-rewards for processing offloaded tasks, meanwhile adapting to the highly dynamic energy harvesting (EH) process and computation task arrivals. RoA is designed based on Lyapunov optimization and Vickrey-Clarke-Groves auction, the operation of which does not require a prior knowledge of the energy harvesting, task arrivals, or wireless channel statistics. Our analytical results confirm the optimality of tasks assignment. Furthermore, simulation results validate the analytical analysis, and verify the efficacy of the proposed RoA.

Heterogeneous networks (HetNets) have been widely accepted as a promising architecture to fulfill the ever-increasing demand for capacity expansion. However, the energy consumed by the dense underlay of the large number of micro base stations that is required to achieve capacity expansion, exacerbates the energy inefficiency of cellular networks. Hybrid energy sources, i.e., the grid and green energy sources, can be used to meet the HetNets excessive demand for energy. In such networks, traffic load balancing becomes crucial to balance the trade-off between green energy utilization and quality of service (QoS) provisioning. Leveraging software-defined radio access networks (SoftRAN) and considering inaccuracy of vital network measurements, we develop an autonomous, robust and resilient load balancing framework. The framework consists of two major modules. First, the H∞regulator module, which guides the temporal utilization of green energy and distribution of network loads among base stations (BSs) in order to achieve long-term average QoS provisioning. Second, a user association module that optimizes user association and its corresponding traffic loads to minimize the network traffic latency while respecting loads proposed by the H∞regulator. Extensive performance evaluations demonstrate the efficacy of the proposed framework in autonomously balancing the trade-off between green energy consumption and traffic latency. Furthermore, performance evaluations confirm the robustness of the proposed framework to estimation inaccuracy and its resilience to sudden changes in network parameters.

The exponential growth of the Information and Communication Technology (ICT) sector have led to a significant increase in energy consumption, higher electricity bills, and negative environmental and economical impacts. Several researchers, network providers, and manufacturers have been investigating different approaches to improve the energy efficiency of communication networks. Software- defined Networks (SDN) is emerging as a new networking framework that separates data plane from control plane in order to simplify network management, reduce operational costs (OPEX), and facilitate innovation. In this work, we address the centralized integral routing problem in SDN. We propose a greedy heuristic algorithm called Energy Efficient Integral Routing (EEIR) algorithm to minimize power consumption in SDN backbone networks while respecting discreteness of link rates. The performance of EEIR has been evaluated in real topologies, and compared to both optimal and shortest path solutions. Experimental results have shown a significant power saving that is as large as 44.42% can be achieved. Compared to optimal solution, EEIR provides a solution with an optimality gap in the range 7.52% - 12.67%.

Due to the limited battery power of sensor nodes and harsh deployment environment, it is of fundamental importance and a great challenge to achieve high energy efficiency and strong robustness in large-scale wireless sensor networks (LS-WSNs). To this end, we propose two self-organizing schemes for LS-WSNs. The first scheme is the energy-aware common neighbor scheme, which considers the neighborhood overlap in link establishment. The second scheme is energy-aware low potential-degree common neighbor (ELDCN) scheme, which considers both neighborhood overlap in topology formation and the potential degrees of common neighbors. Both schemes generate clustering-based and scale-free-inspired LS-WSNs, which are energy-efficient and robust. However, the ELDCN scheme shows higher energy efficiency and stronger robustness to node failures, because it avoids establishing links to hub-nodes with high potential connectivity. Analytical and simulation results demonstrate that our proposed schemes outperform the existing scale-free evolution models in terms of energy efficiency and robustness.

In this paper, we study resource management and allocation for energy harvesting cognitive radio sensor networks (EHCRSNs). In these networks, energy harvesting supplies the network with a continual source of energy to facilitate the self-sustainability of the power-limited sensors. Furthermore, cognitive radio enables access to the underutilized licensed spectrum to mitigate the spectrum-scarcity problem in the unlicensed band. We develop an aggregate network utility optimization framework for the design of an online energy management, spectrum management, and resource allocation algorithm based on Lyapunov optimization. The framework captures three stochastic processes: energy harvesting dynamics, inaccuracy of channel occupancy information, and channel fading. However, a priori knowledge of any of these processes statistics is not required. Based on the framework, we propose an online algorithm to achieve two major goals: first, balancing sensors' energy consumption and energy harvesting while stabilizing their data and energy queues; second, optimizing the utilization of the licensed spectrum while maintaining a tolerable collision rate between the licensed subscriber and unlicensed sensors. The performance analysis shows that the proposed algorithm achieves a close-to-optimal aggregate network utility while guaranteeing bounded data and energy queue occupancy. The extensive simulations are conducted to verify the effectiveness of the proposed algorithm and the impact of various network parameters on its performance.

In this paper, we present three network evolution models for generating fault-tolerant and energy- efficient large-scale peer-to-peer wireless sensor networks (WSNs) based on complex networks theory. Being scale-free is one of the intrinsic features of complex networks-based evolution models that generates fault- tolerant topologies. In this work, we argue that fault- tolerant topologies are not necessarily energy efficient. The three proposed energy-aware evolution models are energy-aware common neighbors (ECN), energy- aware large degree promoted (ELDP) and energy-aware large degree demoted (ELDD). ECN considers neighborhood overlap, whereas ELDP and ELDD consider topological overlap for node attachment. The ELDP model promotes the establishment of links to nodes with a large degree, whereas the ELDD model demotes this strategy. Performance evaluations demonstrate that the proposed models outperform a candidate clustering-based model, thereby providing greater energy savings and fault- tolerance. Among the proposed models, ECN is the winner in-terms of energy efficiency, ELDD performs best in- terms of fault-tolerance, and ELDP conveniently provides balance between the two.

Fair weights have been implemented to maintain fairness in recent resource allocation schemes. However, designing fair weights for multiservice wireless networks is not trivial because users' rate requirements are heterogeneous and their channel gains are variable. In this paper, we design fair weights for opportunistic scheduling of heterogeneous traffic in orthogonal frequency division multiple access (OFDMA) networks. The fair weights determine each user's share of rate for maintaining a utility notion of fairness. We then present a scheduling scheme which enforces users' long term average transmission rates to be proportional to the fair weights. The proposed scheduler takes the advantage of users' channel state information and the inherent flexibility of OFDMA resource allocation for efficient resource utilization. Furthermore, using the fair weights allows flexibility for realization of different scheduling schemes which accommodate a variety of requirements in terms of heterogeneous traffic types and user mobility. Simulation based performance analysis is presented to demonstrate efficacy of the proposed solution in this paper.

In this paper, a novel Kalman filter-based power allocation scheme is developed for cooperative networks with inaccurate channel state information (CSI). The channel estimation error is embedded in the power allocation model which results in uncertain linear and time varying system. The robust and constrained Kalman filter (RCKF) scheme adapts the allocated power to channel variation while satisfying the average bit error probability (BEP) requirements of each subscriber. The proposed scheme's low complexity and robustness to channel estimation inaccuracy make it practical for real networks implementations. Simulation results show that the proposed scheme converges to the optimal allocation with light computational burden despite of the inaccuracy in the received CSI.

Maintaining fairness using weighting factors is a common approach in resource allocation. However, computing weighting factors for multiservice wireless networks is not trivial because users' rate requirements are heterogeneous and their channel gains are variable. In this paper, we propose weighting factor computation and scheduling schemes for orthogonal frequency division multiple access (OFDMA) networks. The weighting factor computation scheme determines each user's share of rate for maintaining a utility notion of fairness. We then present a scheduling scheme which takes the users' weighting factors into consideration to allocate sub-carriers and power in OFDMA networks. The simulation results demonstrate that the proposed scheduling scheme outperforms an opportunistic scheme in terms of fairness performance in different scenarios, where the users are fixed or mobile.

This paper addresses practical implementation issues of resource allocation in OFDM A networks: inaccuracy of channel state information (CSI) available to the resource allocation unit (RAU) and diversity of subscribers' quality of service (QoS) requirements. The resource allocation problem in the considered point-to-multipoint (PMP) network is modeled as a network utility maximization (NUM) problem that allocates subcarriers, rate and power while satisfying orthogonal frequency division multiple access (OFDMA) constraints and QoS constraints defined in the service level agreement. Performance evaluation findings support our theoretical claims: a substantial data rate gain is achieved by considering the CSI imperfection and multiservice classes are supported with QoS guarantees by coordinating with a call admission control (CAC) scheme.

This paper analyzes the ergodic mutual information of OFDMA cooperative relay networks when the Channel State Information (CSI) imperfection is considered at resource allocation unit. Based on the derived ergodic mutual information, a quantitative characterization of the impact of CSI inaccuracy on the ergodic mutual information is presented. A network with multiple subscriber stations, relay stations and one base station that employ a Selection-Decode-and-Forward (SDF) relaying scheme is considered in the uplink mode. Subscriber stations act as relay stations to assist other transmitting subscriber stations. Numerical evaluations illustrate that considering the CSI imperfection based on priori knowledge of the error statistics brings substantial gain to the network in terms of ergodic mutual information, and takes the MAC layer resource allocation algorithms a step ahead towards practical implementations.

In this paper, we focus on the resources allocation for the OFDMA based two-hop relay network which consists of a single base station, dedicated fixed relay stations and subscriber stations. Subscriber stations are allocated the subcarriers and relay stations that are required to satisfy their minimum rate requirements in either non-cooperative mode (i.e., direct communication with the base station) or in cooperative mode with one of the available relay stations. The cooperation is limited to one relay station to reduce the complexity incurred by the need for synchronization with multiple relays and with the base station at the PHY layer. The subcarriers and relay stations allocation problem is formulated as a Binary Integer Programming (BIP) problem with QoS constraints (minimum rate) and a practical synchronization constraint (cooperation with a single relay). Since the formulated problem is NP-complete, a simple sub-optimal algorithm is proposed to manage the multi-service network resources. Simulations and complexity analysis show that the presented algorithm achieves a network near optimal resources allocation with low computational complexity.