Distributed Slice Selection-Based Computation Offloading for Intelligent Vehicular Networks

Abstract

Distributed artificial intelligence (AI) is becoming an efficient approach to fulfill the high and diverse requirements for future vehicular networks. However, distributed intelligence tasks generated by vehicles often require diverse resources. A customized resource provision scheme is required to improve the utilization of multi-dimensional resources. In this work, a slice selection-based online offloading (SSOO) algorithm is proposed for distributed intelligence in future vehicular networks. First, the response time and energy consumption are reduced for processing tasks locally on the vehicles. Then, the offloading overheads, including latency and energy consumption, are calculated by considering the available resource amount, wireless channel states and vehicle conditions. The slice selection results is obtained by the deep reinforcement learning (DRL)-based method. Based on the selection solution, resource allocation results are achieved by KKT conditions and bisection method. Finally, the experimental results depict that the proposed SSOO algorithm outperforms other comparing algorithms in terms of energy consumption and task completion rate.

Federated Fuzzy Neural Network With Evolutionary Rule Learning

“…Recently, personalized FL (PFL) [19], [20], [32], [33] has been proposed; this approach aims to process heterogeneous local data with personalized models. Many of the existing PFL methods were proposed to solve the distributed meta-learning problem [33]- [36]. Among the methods that target normal PL tasks, multitask learning [37] is applied to learn personalized local models by treating each client as a learning task; model mixing [38], [39] achieves the same goal by allowing clients to learn a mixture of the global model and local models.…”

Machine Learning in Network Slicing—A Survey

“…Their results show that their proposed approach outperforms the fixed resource slicing scheme in terms of RB utilization, latency, data rate, and service outage. Moreover, paying particular attention to the vehicular user equipment (VUE), the works in [143] and [144] study RB allocation with respect to multiple slices, by accounting for the channel gain of the vehicular network environment. In this regard, they utilize DNNs to extract features from non-linear relationships among VUEs, thereby finding the optimal resource assignment policies.…”

“…In this regard, they utilize DNNs to extract features from non-linear relationships among VUEs, thereby finding the optimal resource assignment policies. It is learned from their numerical results that having a good knowledge of channel gain can reduce the radio resources overhead cost to 50% [143] and yet energy efficiency is approximately 26% better than baselines in a certain scenario [144].…”

Machine Learning in Network Slicing - A Survey

“…Their results show that their proposed approach outperforms the fixed resource slicing scheme in terms of RB utilization, latency, data rate, and service outage. Moreover, paying particular attention to the vehicular user equipment (VUE), the works in [161] and [162] study RB allocation with respect to multiple slices, by accounting for the channel gain of the vehicular network environment. In this regard, they utilize DNNs to extract features from nonlinear relationships among VUEs, thereby finding the optimal resource assignment policies.…”

“…In this regard, they utilize DNNs to extract features from nonlinear relationships among VUEs, thereby finding the optimal resource assignment policies. It is learned from their numerical results that having a good knowledge of channel gain can reduce the radio resources overhead cost to 50% [161] and yet energy efficiency is approximately 26% better than baselines in a certain scenario [162].…”