Privacy-preserving and truthful auction-based resource allocation mechanisms for task offloading in mobile edge computing

Dongkuo Wu, Xingwei Wang, Xueyi Wang, Siyu Yan, Qiang He, Min Huang

Abstract: The proliferation of mobile devices (MDs) has accelerated the rapid development of various Internet of Things (IoT) applications, which are seeking efficient task execution paradigms. Following this trend, mobile edge computing (MEC) has recently emerged as a promising paradigm that provides offloading services to MDs at the edge of radio access networks. Since the computation and communication resources of edge service providers (ESPs) are limited, effective resource allocation mechanisms are crucial for realizing the MEC paradigm, enabling ESPs to collaborate and provide resources across JointCloud. However, existing works overlook the dependencies among tasks in IoT applications and their associated quality of service (QoS) requirements, while also failing to account for the potential privacy leakage that MDs may suffer from inference attacks. To overcome the above issues, we propose a privacy-preserving and truthful auction-based resource allocation mechanism for task offloading (PRATO) in an MEC system. Specifically, we first design a dependency-aware task offloading strategy determination algorithm, aiming to acquire the resource profiles and the corresponding task offloading solutions of MDs under their QoS requirements. Building on this algorithm, we further design the differentially private winner determination and pricing algorithm to determine the unit payment prices of combinatorial resources as well as the set of winning MDs. Strict theoretical analysis proves that PRATO achieves differential privacy, individual rationality, computational efficiency, truthfulness, and approximate revenue maximization. Extensive simulation results also validate the effectiveness of PRATO.