Approximate and delayed computing have emerged as promising paradigms to offer flexibility in computational accuracy and strategically differentiate tasks between edge and cloud execution to enhance resource utilization. However, these approaches introduce tradeoffs, potentially compromising accuracy on one hand and increasing latency on the other. In this paper, we explore the integration of approximate and delayed computing paradigms within the edge-cloud computing continuum. Users can either offload tasks for approximate computing at the edge or opt for exact but potentially delayed computing at the cloud. In this context, the joint problem of computation task offloading and data compression is formulated and solved as a non-cooperative Game in Satisfaction Form. Each user autonomously determines the amount of task to offload for either computing option and the percentage of data compression for approximate computing, aiming to achieve an acceptable accuracy-latency tradeoff. The formulated game admits a Satisfaction Equilibrium (SE) point, which is concluded using a Reinforcement Learning (RL)-based algorithm. Simulation results demonstrate the performance of the proposed task offloading framework in the achieved accuracylatency tradeoff compared against different offloading strategies.

Collaborative Machine Learning (ML) paradigms, such as Federated Learning (FL), suffer from unequal client contributions and adversarial behavior, where clients deliberately degrade global model accuracy via outdated or poisoned updates. In this paper, we address fair client collaboration and adversarial behavior detection and mitigation using a combined reputation-aware incentive and robust aggregation approach. First, the long-term client reputation across FL epochs is estimated using a variant of the Shapley value, which offers polynomial complexity, contrariwise to the latter. Client reputation is then used to weight model updates during global model generation, effectively mitigating the impact of model poisoning and replay attacks. Moreover, it is used to determine informed monetary rewards for the clients on the server side, which, in turn, guide their efforts in the number of model iterations employed during local training. The interplay between server and clients in their reward and effort decisions is modeled as a Stackelberg game, which concludes fair client participation. Numerical results via modeling and simulation validate the effectiveness of both mechanisms against state-of-the-art and baseline alternatives.

Though extensive research efforts have been devoted to the problem of computing resource pricing, they mainly focus on single computing paradigms. In this paper, we provide a holistic approach to this problem, by treating the whole computing continuum, consisting of cloud, edge, and fog computing providers, simultaneously offering their resources to the users. Within such a complex setting, we establish the concept of symbiotic computing resource pricing and sharing, where the computing providers and the users coexist within a mutually beneficial ecosystem, sharing services and resources as a means of ensuring their business survival and service satisfaction. Under this prism, we introduce two key pricing families, namely the non-cooperative one which involves competition and is treated through game theoretic approaches, and the cooperative resource pricing (full or partial), which addresses complex scenarios through optimization and coalition. A thorough performance assessment is provided, through modeling and simulation, in order to highlight and quantify the key characteristics and tradeoffs of the various resource pricing approaches introduced.

Crowdsourcing plays a critical role in modern information gathering and task execution, yet it faces challenges regarding the task selection and equitable monetary incentives distribution. In this paper, we introduce the TOPMG framework, which addresses these challenges by enabling the workers to select tasks based on their historically experienced monetary incentives and the platforms’ trustworthiness. Specifically, the TOPMG framework utilizes a reinforcement learning approach based on the principles of Optimistic Q-learning with Upper Confidence Bound (OQ-UCB) algorithm, guiding the platform selection process by considering the workers’ monetary incentives, profit, and the platforms’ trustworthiness. Also, the proposed framework introduces a multilateral bargaining game to allocate the platforms’ monetary incentives to the workers by prioritizing their information contribution, fairness, and the platforms’ reputation. Simulation results demonstrate TOPMG’s operational dynamics, scalability, and efficacy, as well as its superiority over existing methodologies.

The efficient operation of the unified Integrated Sensing and Communication (ISAC) – Mobile Edge Computing (MEC) systems is important for enhancing data sensing, communication, and computation processes in next-generation wireless systems. Despite prior research focusing on these systems, little attention has been given to optimizing the device-edge server associations. This paper addresses this gap by introducing the novel two-stage device-edge server association Synergia framework. Firstly, representative utility functions capture the characteristics of the devices and MEC servers by jointly considering their sensing, communication, and computation characteristics. Secondly, the Estimated Synergia framework leverages the Matching Theory to rapidly determine an initial device-server matching by disregarding the devices’ externalities, i.e., the matching decisions of other devices. Thirdly, the Accurate Synergia model refines and improves this matching by using the coalition formation games, while considering the devices’ externalities in optimizing the utilities of both the devices and the MEC servers. Extensive numerical evaluations demonstrate the Synergia’s operational efficiency and scalability, outperforming reinforcement learningbased approaches. Also, a real-world application involving car accident detection validates its applicability.

The Information-Centric Networking (ICN) paradigm has reshaped the modern network architectures and promises efficient content delivery to the end-users. This paper introduces TRUSTCACHE, a novel framework enabling the content caching within ICNs and focusing on the trust-based ICN selection and optimal cache memory allocation to the Content Providers (CPs). The TRUSTCACHE framework incorporates the Optimistic Q-learning with Upper Confidence Bound reinforcement learning algorithm that enables the CPs to autonomously select ICNs based on their cache memory availability and trust levels. Also, TRUSTCACHE enables the CPs to jointly consider the reliability of the ICNs and their cache memory availability by integrating a novel trust model. Furthermore, TRUSTCACHE leverages the multilateral bargaining principles in order to ensure the optimal cache memory allocation among the CPs, in terms of aligning with their profit margin characteristics. Simulation-based experiments validate TRUSTCACHE’s operational efficiency across diverse CP profit margin profiles and highlight its superiority over alternative models lacking trust-based ICN selection or employing proportional fairness strategies for cache memory allocation.

Multi-provider multi-user multi-access edge computing provides a recent market-driven networking paradigm facilitating the user data offloading process. In this paper we introduce the AGORA framework, which employs a sophisticated multi-leader multi-follower Stackelberg game that jointly optimizes the data offloading, computing resource allocation, and computing resource pricing, all facilitated through a non-cooperative game-theoretic approach. In order to support the aforementioned modeling and approach, a novel utility function that quantifies the users satisfaction, factoring in the computing service cost, and an innovative profit function for the MEC providers is introduced, emphasizing the market penetration and the computing service provision costs. Numerical results, obtained via modeling and simulation, demonstrate AGORA’s remarkable adaptability, accommodating homogeneous and heterogeneous user computing demands, while simultaneously outperforming proportional fairness resource allocation approaches, and significantly enhancing the MEC providers’ profitability and the users’ satisfaction from the edge computing services.

Information Centric Networking (ICN) paradigm exploits the in-network caching capacity to support the process of fast and efficient content distribution. In addition to the algorithmic and implementation challenges associated with the decision-making of content placement, the sustainability of content caching frameworks heavily depends on the design of appropriate network economics models to define and support the interactions among the involved players. In order to treat this need, in this paper, considering multiple Content Providers (CPs) while exploiting the in-network caching model, we particularly examine the joint problem of maximizing the CPs profit and their market penetration in terms of attracting a large portion of customers. The problem is formulated as a non-cooperative game among the CPs and the existence and uniqueness of a Pure Nash Equilibrium (PNE) are proven. The performance evaluation of the proposed network economics-based approach is achieved via modeling and simulation, while its superiority against other alternatives is demonstrated.

Integrated Sensing and Communication (ISAC) has recently emerged as an additional communication service within the Internet of Things (IoT) and Cyber-Physical Systems (CPS) era, through which distributed nodes are able to communicate their sensing information to a Base Station (BS) using integrated signals. In this paper, we study the coexistence of ISAC with other communication types of the nodes, by introducing a software-defined framework to control the nodes' uplink transmission powers related to each service. Each node is simultaneously engaged in two types of communications with different BSs for ISAC and generic data transmission to the cloud via dual connectivity. The uplink power splitting/control problem between the BSs is formulated as a non-cooperative game in satisfaction form, through which each node autonomously concludes to a Satisfaction Equilibrium (SE) point that meets its minimum ISAC and pure communication-oriented requirements. Different achievable SE points are analyzed, while an Reinforcement Learning (RL) and a searching-based algorithm are introduced to conclude to the SE and Minimum Efficient SE (MESE) of the studied problem. Simulation results demonstrate the operation of the algorithms and the overall proposed framework in achieving an efficient share of the resources to the different services.

In this paper, we consider a competitive aerial computation offloading environment, where two edge resource operators provide computing services on a time-slot basis to multiple users, via Unmanned Aerial Vehicles (UAVs), each bearing a mounted edge server. The aim of each UAV is to selfishly maximize the difference between its personal and the opponent UAV's utility, by competitively allocating its energy resources to the different users in the system. The problem is formulated as a Generalized Colonel Blotto (GCB) game, where the UAVs allocate their resources across a number of battlefields, i.e., the users, as competing players, seeking to win the battlefield by increasing the difference of their in-between allocated resources and thus, experienced utility. The overall framework is complemented by a Reinforcement Learning (RL)-empowered algorithm to account for the energy efficient scheduling of the UAVs' overall available energy in the different time slots, where the GCB game is realized. The performance evaluation of the proposed framework is achieved via modeling and simulation. The obtained numerical results demonstrate the operation of the proposed GCB game, under different levels of competitiveness between the UAVs, and assess the effectiveness and efficiency of the proposed RL algorithm against different comparative scenarios.