Research On Vehicle Localization And Spectrum Allocation Strategy In VANETs

The sustained efforts to apply information technologies to solving transportation problems stimulated the integration of computer networks and transportation systems. Intelligent Transportation Systems (ITS), as the paradigm of the combination, has advanced with exceptional speed. Vehicular Ad Hoc Networks (VANETs), as the most important part of ITS, has received considerable research interest. In VANETs, vehicles are equipped with wireless communication devices that allow them to exchange information with each other in Vehicle-to-Vehicle communication (V2V) and also to exchange information with roadside network units in Vehicle-to-Roadside communication (V2R). Many applications can be deployed on VANETs, such as vehicle collision warning, driver assistance, dissemination of traffic information, map navigation, automatic parking and Internet access.Almost all of these applications deployed in VANETs require the support of localization system. Usually, localization service is done using Global Positioning System (GPS) and Geographic Information System. Unfortunately, GPS receivers are not the best solution in VANETs, since it cannot work in indoor or dense urban areas where there is no Line of Sight (LOS) to at least four satellites. Therefore, GPS information is likely to be combined with other localization techniques.One of the most important aspects of improving performance of VANET is the high speed of the vehicles. In this case, the mutual wireless communication window is very short due to a relatively small transmission range which can significantly degrade the throughput. Therefore, how to elevate the capability of simultaneously transmitting in VANETs is the chief consideration. If equipped with multiple wireless interfaces, each vehicle can simultaneously transmit over different channels, which has positive impact on the extension of communication window and the promotion of network throughput. Challenging research issues in multi-interface vehicular ad hoc networks are the spectrum allocation and the routing problems.Aim at the above-mentioned problems, the research is mainly focused on the following aspects:(1) Local relative position maps can be determined by vehicles by measuring the distances between its neighbors and exchanging the distance information with other vehicles in multihop communication. Most existing vehicle localization schemes only consider static estimating problem, while little or no attention is paid to the effect of vehicle movement on localization performance. We propose a real-timely VANET localization algorithm based on least square optimization. In our approach, the time is divided into short discrete intervals, so that the vehicle localization problem can be reduced to a convex constraint optimization problem. Unfortunately, the convergence rate of least square is often slow. Therefore, an improved algorithm based on gradient search is proposed.(2) Consider that the former two algorithms need to construct complete distance matrix, we propose a semi-definite programming (SDP) localization algorithm for VANETs based on the convex constraint optimization model. SDP takes full advantage of convex constraints of distances between vehicles that yet to be localized. Furthermore, gradient descent optimization can be incorporated into our algorithm to further improve the estimating accuracy at the expense of additional cost.(3) Aiming at the lower utilization of spectrum resource in vehicular networks, an algorithm base on semidefinite programming is presented, which can be utilized to dynamic spectrum allocation in multi-channel vehicular networks. Firstly, the algorithm needs to measure the distances among the surrounding vehicles to compute conflict graph. And then the semidefinite programming relaxation is employed to coordinate the competition of spectrum resource. Finally, the ultimate spectrum assignment is computed based on the initial channel assignment and compression process.(4) A strategy of spectrum allocation and geo-routing base on clustering is presented, which could be utilized to multi-interface vehicular networks. Firstly, the network was clustered according to the velocity of vehicles. And then an off-line spectrum allocation scheme was applied to the inter-cluster links, thus eliminating the impact of the mobility of the vehicles. Because the velocity of the vehicles coming from same cluster is similar, the topology of intra-cluster is stable which is helpful to the application of dynamic spectrum allocation technology.In summation, we deal with vehicle self-localization preblem, which can serve as a major supplement for GPS navigation. Geographic-based dynamic spectrum allocation in VANETs is another research topic in this paper. Multi-interface spectrum allocation technology has been shown to be an effective measure to increase the performance of VANETs.