Matching game theoretical model for stable relay selection in a UAV-assisted internet of vehicles

Abstract

This paper tackles the problem of selecting stable relays for Optimized Link State Routing protocol (OLSR) in urban Internet of Vehicles (IoV) in the presence of Unmanned Aerial Vehicles (UAVs). With the evolution of Internet of Things (IoT), IoV emerged from the conventional vehicular ad-hoc network to enable Vehicle to Everything (V2X) communication through different routing protocols. In OLSR protocol, Multi-Point Relays (MPRs) are selected based on their reachability and uniqueness to route information. However, urban environments are characterized by the rapid changes in topology due to the presence of intersections and traffic lights. Although clusters were proposed as a solution, selecting stable heads and MPRs are of significance, considering environment metrics to provide higher connectivity. As a solution to improve routing in IoV, a distributed Gale-Shapley matching game is proposed for stable clustering and MPR selection, utilizing nodes' quality of service (QoS). Nodes' QoS is calculated using Bayesian Belief function of stable connections utilizing environment metrics. For further enhancement in the network performance, UAVs are integrated in the network of vehicles. Conducted simulations show high percentage of stable heads and MPRs for the proposed model compared to benchmark protocols. In addition, the proposed model shows high performance in terms of packet delivery ratio, throughput and End-to-End delay, which are further improved by the presence of UAVs.

Introduction

Background. Internet of Vehicles (IoV) emerged from the conventional vehicular ad-hoc network (VANET) as part of Internet of things (IoT) [1]. In contrast to VANET, IoV is characterized by a high degree of manageability and ope-rationalization including intelligent vehicles that provide potential services to users. Several applications have been developed for IoV that can be categorized into safety and user applications including collision avoidance, driving assistance, multimedia streaming and global internet access [1]. To share data and provide services to users, routing protocols for the complex network of IoV imply several communication links such as Vehicle-to-Vehicle (V2V), Vehicle-to-Infrastructure (V2I) and Vehicle-to-Pedestrian (V2P), that can be generalized as Vehicle-to-everything (V2X). Routing in urban environment is considered a challenge in IoV, as a result of rapid changes in topology. Hence, data routes are continuously disconnected resulting in high drops of data packets [1].

Motivation. Intersections and traffic lights are main characteristics of urban environments that contribute to the dynamic nature of vehicular networks [2]. Therefore, several studies were considering environment metrics in proposing VANET routing protocols and relay selection [3], [4], [5]. Although street-centric QoS-OLSR routing protocols proposed in [3] and [5] provide adequate network performance, they do not consider the link quality in the selection of relays as proposed in [4]. In [6], it was found that the proactive routing protocol, Optimized Link state Routing (OLSR), is more adequate for vehicular communication in urban environment, in contrast to the latter studies which have developed reactive and position-based routing protocols. [7] and [8] proposed routing protocols by enhancing the Multi-Point Relays (MPRs) selection in OLSR using environment metrics and mobility of vehicles. Although these routing protocols consider environment metrics in relay selection, they neglect the selection of stable relays, which are the nodes with low probability of disconnecting from their neighbors. Selecting stable relays for routing decreases the rapid changes in topology, which enhances the network performance [9].

Game theory was also exploited in IoV for relay selection as proposed in [2], in addition to clusters and coalition formation as proposed in [9] and [10]. The implementation of game theory in [9] requires calculating a utility function which needs global information that is managed and shared by a centralized node. This can be seen as a limitation of game theory in a distributive network [11]. Matching theory has been introduced in IoV in many researches recently as a more practical alternative solution to game theory. Many characteristics distinguish matching theory for two distinct sets in IoV such as ensuring network stability, defining the preferences of players to impose system constraints and simple and efficient interaction between players for distributive model [11]. Therefore, several applications have utilized matching theory in IoV for data sharing, caching and edge computing as proposed in [11], [12] and [13]. However, to the best of our knowledge, matching theory was not exploited in stable relay selection in IoV, although it promotes stability.

As the number of Unmanned Aerial Vehicles (UAVs) utilized in different applications, commercial, civilian and scientific, are increasing, they are also utilized to act as relays for IoV routing protocols [14]. UAVs are equipped with wireless communication transceivers, in addition to their ability to avoid obstacles and provide line of sight communication, making them suitable nodes to be selected as relays [14], [15]. As an alternative solution to Base Stations (BSs), UAVs are cheaper to deploy and can provide higher coverage for the ground area [16]. Several studies exploited the use of UAVs as relays in IoV to enhance the network performance [16], [17], [18], [14], [19]. Although connectivity metrics such as the connection duration were considered in the UAV-assisted routing protocols in the related work, selecting stable relays were neglected. Similar to vehicles, the random or constrained mobility of UAVs adds to the rapid changes in topology which lead to frequent disconnections in the routing paths.

Problem. This paper addresses the problem of selecting stable relays for OLSR in a UAV-assisted urban vehicular network. While the mobility of vehicles is a reason for rapid changes in topology, urban environments characteristics, such as intersections and traffic lights, play a role in the dynamic network. When vehicles approach the intersections, they may disconnect depending on their lane and mobility. In addition, traffic lights state affects the connectivity of vehicles as they may stop at the intersections or leave in the same or different directions. Although, clusters were proposed to decrease the instability of topology [9], [11], selecting unstable heads or MPRs results in frequent re-formation of clusters. Therefore, stable relays should be selected with high probability of keeping the connections with their neighbors after leaving the intersections. Also, in a UAV-assisted network, vehicle to UAV (V2U) connections can be easily broken due to random mobility of UAVs. However, the high coverage of UAVs makes them suitable to act as relays. Therefore, selecting UAVs as stable relays can improve the network performance.

The main challenges addressed in this work are:

• The selection of stable cluster heads and MPRs.

• The effect of rapid changes in topology as a result of urban environment metrics such as intersections and traffic lights.

• Random mobility of UAVs when selected as stable relays.

Contribution. This paper proposes a many-to-one Gale-Shapley matching game model utilized for clustering and selecting MPR for OLSR in urban IoV to achieve stability. Hence, the network performance is enhanced in terms of packet delivery ratio, throughput, end-to-end delay and overall network stability. The proposed matching game model is divided into two sub-models; head-member matching and head-MPR matching. In the head-member matching game, a stable head is selected by multiple members. Followed by head-MPR matching game, where a repeated version of the many-to-one Gale-Shapley results in matching many heads to many possible MPRs. The main players of the game are the vehicles and UAVs. Players are divided into single match and multi-match nodes. Each node has a quota size defining the size of its matches. Single match nodes have a quota size $q=1$, while, multi-match nodes have a quota size $q\ge 1$. The main objective of the matching game is to select stable relays, whether heads or MPRs, that is imposed by the preferences of the players in the matching game. Preferences of the players are defined based on their Quality of Service (QoS) and reputation. The QoS of a node reflects its stability with respect to its neighbors, while reputation represents node's reliability in acting as relay motivating it to cooperate inspired by [2]. The stability of a node is calculated using a Bayesian belief function describing a node's belief of having stable connections to its neighbors given the node's probability of staying or leaving a street segment or zone.

The main contribution of this work is a Gale-Shapley matching based MPR selection for OLSR protocol in urban IoV that considers:

• The QoS function as a metric of nodes' stability using Bayesian belief function utilizing environment metrics to calculate time to leave street/zone.

• The preferences of single match nodes built based on QoS and reputation, while multi-match nodes' preferences depend on incentives, motivating them to cooperate.

• UAVs as stable relays to improve the network performance.