Multi-Dimensional Resource Orchestration in Vehicular Edge Networks

Abstract

In the era of autonomous vehicles, the advanced technologies of connected vehicle lead to the development of driving-related applications to meet the stringent safety requirements and the infotainment applications to improve passenger experience. Newly developed vehicular applications require high-volume data transmission, accurate sensing data collection, and reliable interaction, imposing substantial constrains on vehicular networks that solely rely on cellular networks to fetch data from the Internet and on-board processors to make driving decisions. To enhance multifarious vehicular applications, Heterogeneous Vehicular Networks (HVNets) have been proposed, in which edge nodes, including base stations and roadside units, can provide network connections, resulting in significantly reduced vehicular communication cost. In addition, caching servers are equipped at the edge nodes, to further alleviate the communication load for backhaul links and reduce data downloading delay. Hence, we aim to orchestrate the multi-dimensional resources, including communication, caching, and sensing resources, in the complex and dynamic vehicular environment to enhance vehicular edge network performance. The main technical issues are: 1) to accommodate the delivery services for both location-based and popular contents, the scheme of caching contents at edge servers should be devised, considering the cooperation of caching servers at different edge nodes, the mobility of vehicles, and the differential requirements of content downloading services; 2) to support the safety message exchange and collective perception services for vehicles, communication and sensing resources are jointly allocated, the decisions of which are coupled due to the resource sharing among different services and neighboring vehicles; and 3) for interaction-intensive service provisioning, e.g., trajectory design, the forwarding resources in core networks are allocated to achieve delay-sensitive packet transmissions between vehicles and management controllers, ensuring the high-quality interactivity. In this thesis, we design the multi-dimensional resource orchestration schemes in the edge assisted HVNets to address the three technical issues.

Firstly, we design a cooperative edge caching scheme to support various vehicular content downloading services, which allows vehicles to fetch one content from multiple caching servers cooperatively. In particular, we consider two types of vehicular content requests, i.e., location-based and popular contents, with different delay requirements. Both types of contents are encoded according to fountain code and cooperatively cached at multiple servers. The proposed scheme can be optimized by finding an optimal cooperative content placement that determines the placing locations and proportions for all contents. To this end, we analyze the upper bound proportion of content caching at a single server and provide the respective theoretical analysis of transmission delay and service cost (including content caching and transmission cost) for both types of contents. We then formulate an optimization problem of cooperative content placement to minimize the overall transmission delay and service cost. As the problem is a multi-objective multi-dimensional multi-choice knapsack one, which is proved to be NP-hard, we devise an ant colony optimization-based scheme to solve the problem and achieve a near-optimal solution. Simulation results are provided to validate the performance of the proposed scheme, including its convergence and optimality of caching, while guaranteeing low transmission delay and service cost.

Secondly, to support the vehicular safety message transmissions, we propose a two-level adaptive resource allocation (TARA) framework. In particular, three types of safety message are considered in urban vehicular networks, i.e., the event-triggered message for urgent condition warning, the periodic message for vehicular status notification, and the message for environmental perception. Roadside units are deployed for network management, and thus messages can be transmitted through either vehicle-to-infrastructure or vehicle-to-vehicle connections. To satisfy the requirements of different message transmissions, the proposed TARA framework consists of a group-level resource reservation module and a vehicle-level resource allocation module. Particularly, the resource reservation module is designed to allocate resources to support different types of message transmission for each vehicle group at the first level, and the group is formed by a set of neighboring vehicles. To learn the implicit relation between the resource demand and message transmission requests, a supervised learning model is devised in the resource reservation module, where to obtain the training data we further propose a sequential resource allocation (SRA) scheme. Based on historical network information, the SRA scheme offline optimizes the allocation of sensing resources (i.e., choosing vehicles to provide perception data) and communication resources. With the resource reservation result for each group, the vehicle-level resource allocation module is then devised to distribute specific resources for each vehicle to satisfy the differential requirements in real time. Extensive simulation results are provided to demonstrate the effectiveness of the proposed TARA framework in terms of the high successful reception ratio and low latency for message transmissions, and the high quality of collective environmental perception.

Thirdly, we investigate forwarding resource sharing scheme to support interaction intensive services in HVNets, especially for the delay-sensitive packet transmission between vehicles and management controllers. A learning-based proactive resource sharing scheme is proposed for core communication networks, where the available forwarding resources at a switch are proactively allocated to the traffic flows in order to maximize the efficiency of resource utilization with delay satisfaction. The resource sharing scheme consists of two joint modules: estimation of resource demands and allocation of available resources. For service provisioning, resource demand of each traffic flow is estimated based on the predicted packet arrival rate. Considering the distinct features of each traffic flow, a linear regression scheme is developed for resource demand estimation, utilizing the mapping relation between traffic flow status and required resources, upon which a network switch makes decision on allocating available resources for delay satisfaction and efficient resource utilization. To learn the implicit relation between the allocated resources and delay, a multi-armed bandit learning-based resource sharing scheme is proposed, which enables fast resource sharing adjustment to traffic arrival dynamics. The proposed scheme is proved to be asymptotically approaching the optimal strategy, with polynomial time complexity. Extensive simulation results are presented to demonstrate the effectiveness of the proposed resource sharing scheme in terms of delay satisfaction, traffic adaptiveness, and resource sharing gain.

In summary, we have investigated the cooperative caching placement for content downloading services, joint communication and sensing resource allocation for safety message transmissions, and forwarding resource sharing scheme in core networks for interaction intensive services. The schemes developed in the thesis should provide practical and efficient solutions to manage the multi-dimensional resources in vehicular networks.