Mobility-Aware Robust Resource Allocation and Computation Migration in Heterogeneous Virtualized Vehicular Networks by Low Overhead Multi-Agent DRL-based Approach

Citation Author(s):
Mohammad Reza
Mohammad Reza
A. Jorswieck
Submitted by:
Mohsen Pourghasemian
Last updated:
Sat, 12/31/2022 - 18:30
0 ratings - Please login to submit your rating.


In emerging vehicular networks, delay-sensitive tasks can be processed in real time by offloading to

the edge computing servers. Unlike the legacy scenarios, in this paper a novel data offloading/delivery

decision making framework is proposed, where users have the option to divide their task into several

portions and partially offload their data to a complex multi-access edge computing (MEC) environment,

consisting of several MEC servers located on road side units (RSUs), base stations (BSs), and unused

computing power in other vehicles and pedestrians. This can increase the reliability and the spectral

efficiency, and decrease delay. However, due to high mobility of vehicles, the vehicle-mounted MEC

servers act as a uncertain payoff for the users, while the local computation server (e.g., RSUs, and BSs)

alternatives constitute safe and guaranteed options. Therefore, the offloaded tasks may be migrated to

other MEC servers to provide continuous execution. In addition, users’ high mobility and traffic variation

can result in task offloading demand uncertainty and the imperfect channel state information (I-CSI)

estimation that lead to difficulties in guaranteeing the quality of service (QoS) for each task. In this

paper, we consider a two-time scale resource management scheme for task offloading with fluctuated

traffic demands and I-CSI, aiming at minimizing total network cost and delay. We consider a hierarchical

control architecture in our proposed scheme including large time-scale RSU activation/deactivation and

short time-scale resource allocation, and making task offloading decisions. We extend our problem as


multi-time scale Markov decision process. Simulation results validate that our proposed task offloading

scheme can increase the task completion rate by 27% and 98% when considering task partitioning and

task migration, respectively. Moreover, considering demand uncertainty, we can achieve more robust

solution by increasing the task completion rate by 42.8%. Finally, we show that considering MADDPG


can increase task offloading cost by 45% due to communication overhead between agents.



Tensorflow >= 1.15
Keras >= 2.4

Python >= 3.7

After running the "" file, the rewards.txt is created. Simply plot the reward.