[Paper Review] Computation Rate Maximization for Wireless Powered Mobile-Edge Computing with Binary Computation Offloading

Suzhi Bi, Ying Jun|arXiv (Cornell University)|Aug 29, 2017

Energy Harvesting in Wireless Networks21 citations

TL;DR

This paper proposes a joint optimization framework for maximizing computation rate in a wireless-powered mobile-edge computing network with binary offloading, where devices either compute locally or offload tasks entirely to a MEC server. It introduces two algorithms—coordinate descent and ADMM-based decomposition—enabling efficient resource allocation across time, power, and computing mode, with the ADMM method scaling favorably to large networks and achieving near-optimal performance in simulations.

ABSTRACT

In this paper, we consider a multi-user mobile edge computing (MEC) network powered by wireless power transfer (WPT), where each energy-harvesting WD follows a binary computation offloading policy, i.e., data set of a task has to be executed as a whole either locally or remotely at the MEC server via task offloading. In particular, we are interested in maximizing the (weighted) sum computation rate of all the WDs in the network by jointly optimizing the individual computing mode selection (i.e., local computing or offloading) and the system transmission time allocation (on WPT and task offloading). The major difficulty lies in the combinatorial nature of multi-user computing mode selection and its strong coupling with transmission time allocation. To tackle this problem, we first consider a decoupled optimization, where we assume that the mode selection is given and propose a simple bi-section search algorithm to obtain the conditional optimal time allocation. On top of that, a coordinate descent method is devised to optimize the mode selection. The method is simple in implementation but may suffer from high computational complexity in a large-size network. To address this problem, we further propose a joint optimization method based on the ADMM (alternating direction method of multipliers) decomposition technique, which enjoys much slower increase of computational complexity as the networks size increases. Extensive simulations show that both the proposed methods can efficiently achieve near-optimal performance under various network setups, and significantly outperform the other representative benchmark methods considered.

Motivation & Objective

• To address the dual limitations of finite battery life and low computing capability in IoT and WSN devices.
• To maximize the weighted sum computation rate in a multi-user wireless-powered MEC system with binary offloading.
• To jointly optimize computing mode selection (local vs. offload) and transmission time allocation for WPT and offloading.
• To develop low-complexity, scalable algorithms that handle the combinatorial nature of mode selection and coupling with time allocation.
• To evaluate performance across diverse network setups and compare with benchmark methods.

Proposed method

• Proposes a decoupled optimization approach where computing mode selection is fixed first, followed by bi-section search for optimal time allocation under that mode.
• Employs a coordinate descent method to iteratively optimize mode selection, using conditional optimal time allocation as a subroutine.
• Develops an ADMM-based joint optimization method to decompose the problem, enabling scalable computation with O(N) complexity in network size N.
• Uses Lagrangian relaxation and dual decomposition to handle constraints, with the dual variable ν updated via subgradient methods.
• Derives closed-form expressions for optimal time allocation using the Lambert-W function, based on partial derivatives of the Lagrangian.
• Solves the dual problem by finding the unique ν* that satisfies a monotonically decreasing equation Q(ν*)=0, ensuring optimality.

Experimental results

Research questions

• RQ1 How can the computation rate be maximized in a wireless-powered MEC network with binary offloading under joint time and mode optimization?
• RQ2 What is the trade-off between local computation and offloading in terms of energy and latency when devices harvest energy from RF signals?
• RQ3 How can the combinatorial complexity of multi-user mode selection be efficiently managed in large-scale networks?
• RQ4 What is the performance gain of the proposed algorithms compared to conventional benchmarks in terms of computation rate and scalability?
• RQ5 How does the computational complexity of the proposed methods scale with increasing network size?

Key findings

• The coordinate descent (CD) method achieves near-optimal performance but suffers from O(N³) computational complexity, making it less suitable for large networks.
• The ADMM-based method scales linearly with network size, exhibiting O(N) complexity, and is therefore more efficient for large-scale deployments.
• Both proposed algorithms significantly outperform benchmark methods in computation rate across various network configurations and channel conditions.
• The ADMM-based method maintains high performance with much slower complexity growth, making it ideal for massive IoT and dense WSN scenarios.
• Theoretical analysis confirms the existence and uniqueness of the optimal dual variable ν*, ensuring convergence of the optimization process.
• Simulation results validate the effectiveness of the proposed algorithms, showing substantial gains in weighted sum computation rate over conventional schemes.