Distributed task allocation in Mobile Device Cloud exploiting federated learning and subjective logic

Abstract

Mobile Device Cloud (MDC) has become a promising and lucrative cloud environment that exploit nearby mobile devices’ idle resources to improve compute-intensive applications. Computing code at nearby mobile devices rather than a distant master cloud helps improve real-time applications’ performance. However, it is non-trivial to motivate the worker devices to participate voluntarily in sharing their unused resources. In this paper, we have provided a distributed mobile device cloud environment by which workers make their auction decisions distributively and parallelly. We also introduce the federated learning and multi-weight subjective logic-based reputation scheme to measure worker mobile devices’ trustworthiness and reliability. Moreover, a novel utility function for the buyers is proposed considering the cost, Quality-of-Experience (QoE), and the workers’ reputation by which buyers select the most suitable worker in a distributed way. We have also proved that our proposed system achieves the desirable properties of computational efficiency, individual rationality, truthfulness, and budget balance. Empirical evaluations have been carried out in MATLAB that demonstrate the significant performance improvement in terms of QoE and utility of the buyers compared to other state-of-the-art works.