Computing on Wheels: A Deep Reinforcement Learning-Based Approach

Future generation vehicles equipped with modern technologies will impose unprecedented computational demand due to the wide adoption of compute-intensive services stringent latency requirements. The capacity next vehicular networks can be enhanced by incorporating edge or fog computing paradigm. However, growing popularity and massive novel make resources insufficient. A possible solution overcome this challenge is employ onboard computation close vicinity that are not resource-constrained along for enabling tasks offloading service. In paper, we investigate problem task in a practical environment considering mobility electric (EVs). We propose paradigm enables EVs offload their resource hungry either roadside unit (RSU) nearby mobile EVs, which have no restrictions. Hence, formulate non-linear (NLP) minimize energy consumption subject network resources. Then, order solve tackle issue high deep reinforcement learning (DRL) based enable finding best power level communication, an optimal assisting EV pairing, amount required execute task. proposed minimizes overall system pinnacle while meeting requirements posed offloaded Finally, through simulation results, demonstrate performance approach, outperforms baselines terms per consumption.