PageRank-Based Multi-Hop Computation Offloading in D2D Networks

• Published in: IEEE networking letters Vol. 2; no. 4; pp. 195 - 198
• Main Authors: Fan, Wenhao, Teng, Desheng, Wu, Fan, Liu, Yuan'an
• Format: Journal Article
• Language: English
• Published: Piscataway IEEE 01.12.2020; The Institute of Electrical and Electronics Engineers, Inc. (IEEE)
• Subjects: Algorithms; Computation offloading; Computational modeling; D2D networks; Device-to-device communication; Energy consumption; Energy storage; Iterative algorithms; Iterative methods; Markov processes; PageRank; Performance evaluation; Ranking; resource management; Routing; Search engines; Spread spectrum communication; Storage capacity; Task analysis; Weight
• ISSN: 2576-3156; 2576-3156
• DOI: 10.1109/LNET.2020.3037822

In a D2D network, the tasks of a user device (UD) can be offloaded to other UDs to improve the UD's performance. We propose a novel computation offloading scheme to efficiently alleviate the delay of task processing and the energy consumption of UDs. Inspired by PageRank algorithm, the strategies of computation offloading for UDs are determined by their steady ranking values, which reflect the overall influence of UDs' performance in the network. Firstly, we model the network as a directed weighted graph, where the weight of each vertex is obtained by integrating the computing ability, battery energy, and available storage capacity of the corresponding UD, and the weight of each directed edge is obtained by integrating the data rates and potential interference of the D2D connection between the corresponding two UDs. Secondly, a power iteration algorithm, which satisfies Markov convergence, is designed to compute the steady ranking values of UDs. Finally, we design an algorithm to obtain one-hop strategies and multi-hop routing strategies for computation offloading among UDs. The superiority of our scheme is demonstrated in simulations and by comparisons.