Abstract:This paper studies a sequential task offloading problem for a multiuser mobile edge computing (MEC) system. While most of the existing works consider static one-shot offloading optimization with fixed wireless channel conditions and fixed computational tasks, we consider a dynamic optimization approach, which embraces wireless channel fluctuations and random deep neural network (DNN) task arrivals over an infinite horizon. Specifically, we introduce a local CPU workload queue (WD-QSI) and an MEC server workload queue (MEC-QSI) to model the dynamic workload of DNN tasks at each WD and the MEC server, respectively. The transmit power and the partitioning of the local DNN task at each WD are dynamically determined based on the instantaneous channel conditions (to capture the transmission opportunities) and the instantaneous WD-QSI and MEC-QSI (to capture the dynamic urgency of the tasks) to minimize the average latency of the DNN tasks. The joint optimization can be formulated as an ergodic Markov decision process (MDP), in which the optimality condition is characterized by a centralized Bellman equation. However, the brute force solution of the MDP is not viable due to the curse of dimensionality as well as the requirement for knowledge of the global state information. To overcome these issues, we first decompose the MDP into multiple lower dimensional sub-MDPs, each of which can be associated with a WD or the MEC server. Next, we further develop a parametric online Q-learning algorithm, so that each sub-MDP is solved locally at its associated WD or the MEC server. The proposed solution is completely decentralized in the sense that the transmit power for sequential offloading and the DNN task partitioning can be determined based on the local channel state information (CSI) and the local WD-QSI at the WD only. Additionally, no prior knowledge of the distribution of the DNN task arrivals or the channel statistics will be needed for the MEC server. The proposed solution can achieve the superb performance over various state-of-the-art baselines.

Sequential Offloading for Distributed DNN Computation in Multiuser MEC Systems

Joint Computation Offloading and Resource Allocation for MEC-Enabled IoT Systems with Imperfect CSI

Computation Offloading Scheme with D2D for MEC-enabled Cellular Networks

Decentralized DNN Task Partitioning and Offloading Control in MEC Systems With Energy Harvesting Devices

Delay-Optimal Computation Offloading for Computation-Constrained Mobile Edge Networks.

Task Partitioning and Offloading in DNN-Task Enabled Mobile Edge Computing Networks

Task Offloading, Load Balancing, and Resource Allocation in MEC Networks

Dynamic Task Offloading in MEC-Enabled IoT Networks - A Hybrid DDPG-D3QN Approach.

Offloading Strategy for Dependency-Aware Tasks in MEC System Based on Deep Q-network

Dynamic Offloading for Multiuser Muti-CAP MEC Networks: A Deep Reinforcement Learning Approach

Efficient Offloading for Minimizing Task Computation Delay of NOMA-Based Multiaccess Edge Computing

Deep Learning-Based Dynamic Computation Task Offloading for Mobile Edge Computing Networks

Stochastic Computation Offloading and Scheduling Based on Mobile Edge Computing

Cross-User Dependent Task Offloading and Resource Allocation in Spatial-Temporal Dynamic MEC Networks

Online Learning for Distributed Computation Offloading in Wireless Powered Mobile Edge Computing Networks

Decentralized Computation Offloading for Multi-User Mobile Edge Computing: A Deep Reinforcement Learning Approach

Optimized Computation Offloading Performance in Virtual Edge Computing Systems via Deep Reinforcement Learning

Multi-User Computation Offloading with D2D for Mobile Edge Computing

A Multi-Head Ensemble Multi-Task Learning Approach for Dynamical Computation Offloading

AI-Enabled Task Offloading for Improving Quality of Computational Experience in Ultra Dense Networks

A Joint Distributed Computation Offloading and Resource Allocation Scheme for MEC Networks