notesum.ai

Published at November 27

Dependency-Aware CAV Task Scheduling via Diffusion-Based Reinforcement Learning

cs.AI

cs.RO

Released Date: November 27, 2024

Authors: Xiang Cheng¹, Zhi Mao¹, Ying Wang², Wen Wu¹

Aff.: ¹Frontier Research Center, Pengcheng Laboratory, Shenzhen, China; ²School of Information and Communication Engineering, Beijing University of Posts and Telecommunications

Arxiv: http://arxiv.org/abs/2411.18230v1

Parameter	Value
Number of TVs and SVs $(N,S)$	{2, 5}
Number of subtasks of single task $(W)$	{4 $\sim$ 6}
Computing power of TVs, SVs $(f_{n},f_{s})$	{2, 2 $\sim$ 8} GHz
Computing power of BS server $(f_{j})$	50 GHz
Transmit power of vehicle and UAV $(P_{n},P_{u})$	{20, 30} dBm
Bandwidth of vehicle and UAV $(B_{n},B_{u})$	{5, 10} MHz
Maximum tolerant delay of single task $(\tau_{m})$	650 ms
Workload size of a subtask $(\omega_{m}^{w})$	{500 $\sim$ 5000} KB
Length of highway section	1 km
Learning rate and discount factor	{0.001, 0.95}
Experience replay size	100000
Denoising step	10