Multi-dimensional quality-driven service recommendation with privacy-preservation in mobile edge environment

Abstract

With the advance of mobile edge computing (MEC), the number of edge services running on mobile devices grows explosively. In this situation, it is becoming a necessity to recommend the most suitable edge services to a mobile user from massive candidates, based on the historical quality of service (QoS) data. However, historical QoS is a kind of private data for users, which needs to be protected from privacy disclosure. Currently, researchers often use the Locality-Sensitive Hashing (LSH) technique to achieve the goal of privacy-aware recommendations. However, existing LSH-based methods are only applied to the recommendation scenarios with a single QoS dimension (e.g., response time or throughput), without considering the multi-dimensional QoS (e.g., response time and throughput) ensemble, which narrow the application scope of LSH in privacy-preserving recommendations significantly. Considering this drawback, this paper proposes a multi-dimensional quality ensemble-driven recommendation approach named RecLSH-TOPSIS based on LSH and TOPSIS (Technique for Order Preference by Similarity to Ideal Solution) techniques. First, the traditional single-dimensional LSH recommendation approach is extended to be a multi-dimensional one, through which we can obtain a set of candidate services that a user may prefer. Second, we use TOPSIS technique to rank the derived multiple candidate services and return the user an optimal one. At last, a case study is presented to illustrate the feasibility of our proposal to make privacy-preserving edge service recommendations with multiple QoS dimensions.