[Paper Review] Intelligent Reflecting Surface Meets Mobile Edge Computing: Enhancing Wireless Communications for Computation Offloading

Yang Liu, Jun Zhao+5 more|arXiv (Cornell University)|Jan 21, 2020

Advanced Wireless Communication Technologies43 references46 citations

TL;DR

The paper studies computation offloading in IRS-aided wireless networks, proposing feasibility checks and an optimization framework that maximizes edge server earnings while guaranteeing mobile devices’ rate constraints. It shows IRSs can improve offloading feasibility and edge earnings.

ABSTRACT

We consider computation offloading for edge computing in a wireless network equipped with intelligent reflecting surfaces (IRSs). IRS is an emerging technology and has recently received great attention since they can improve the wireless propagation environment in a configurable manner and enhance the connections between mobile devices (MDs) and access points (APs). At this point not many papers consider edge computing in the novel context of wireless communications aided by IRS. In our studied setting, each MD offloads computation tasks to the edge server located at the AP to reduce the associated comprehensive cost, which is a weighted sum of time and energy. The edge server adjusts the IRS to maximize its earning while maintaining MDs' incentives for offloading and guaranteeing each MD a customized information rate. This problem can be formulated into a difficult optimization problem, which has a sum-of-ratio objective function as well as a bunch of nonconvex constraints. To solve this problem, we first develop an iterative evaluation procedure to identify the feasibility of the problem when confronting an arbitrary set of information rate requirement. This method serves as a sufficient condition for the problem being feasible and provides a feasible solution. Based on that we develop an algorithm to optimize the objective function. Our numerical results show that the presence of IRS enables the AP to guarantee higher information rate to all MDs and at the same time improve the earning of the edge server.

Motivation & Objective

• Motivate computation offloading in edge computing to reduce time and energy costs for mobile devices.
• Incorporate intelligent reflecting surfaces to improve uplink rates and network performance.
• Develop a feasibility evaluation and an optimization algorithm to maximize edge server earnings under rate and IRS constraints.
• Provide insight into how IRS-assisted channels affect offloading incentives and payments.
• Demonstrate the benefits of IRS in enabling higher information rates and edge server profits.

Proposed method

• Model a multi-user IRS-aided uplink with edge computing at the AP and single-antenna mobile devices.
• Formulate a utility-based decision framework comparing local versus edge computing under rate constraints.
• Derive an optimization problem to maximize edge server payments subject to rate and IRS amplitude constraints.
• Introduce a feasibility checking approach using SINR/MSE relations to assess rate constraints.
• Apply a simplified problem (P2) to obtain a tractable lower bound and guide the IRS optimization.
• Propose a block coordinate descent based approach to solve the nonconvex IRS optimization problem.

Experimental results

Research questions

• RQ1 Can IRS-aided channels improve the feasibility of meeting information rate constraints for all mobile devices in an edge computing offloading setting?
• RQ2 How should the edge server set payments to maximize its earnings while ensuring mobile devices offload when beneficial?
• RQ3 What is the impact of IRS phase shifts on the achievable rates and offloading decisions in a shared uplink?
• RQ4 Can a tractable optimization formulation provide near-optimal IRS configurations to maximize payments under rate constraints?

Key findings

• IRS-enabled environments increase the feasibility probability of meeting rate constraints for all mobile devices.
• An optimization framework can increase the edge server’s earnings by adjusting IRS reflections and considering payments to offloading devices.
• The proposed feasibility evaluation procedure provides a sufficient condition and a feasible solution when rate requirements are given.
• A simplified problem yields a lower bound on performance and offers practical insights for IRS design in edge computing contexts.
• The results indicate faster convergence of the algorithm and tangible gains in edge server earnings with IRS assistance.