Вы находитесь на странице: 1из 6

EEL6591 Wireless Networks under Class Project under Dr.

Yuguang "Michael" Fang, Spring 2013

Feasibility Analysis and Design of Pseudo-Stationary VANET in absence of Roadside Units


Abhijeet Uike (93619946), University of Florida, abhijeetuike@ufl.edu

Abstract - Ad-Hoc Networks are poised to carry forward the evolutionary baton of human communication systems carried by Cellular Networks in last few decades. Vehicular Ad-Hoc networks promise the first and foremost prominent applications on a wide commercial scale of MANETs. CISCO, NXP, Audi have partnered with COHDA systems to bring 802.11p based VANET communication capability to vehicles on a commercial scale from as early as 2014[5]. However, the process of establishing an entirely new infrastructure for VANETs is going to be a long process and would need support from various bodies. As is the case with developing any new system, VANET establishment is going to face classic chicken-n-egg problem. Early adopters (units in vehicles) will not have complete infra-structure on the roads to communicate which will hinder the service. Whereas lack of more adopters will prevent private entities (like wireless carriers who build networks) from building the roadside infrastructure network. This paper proposes a unique new idea to address this phase of development of VANET infrastructure. Paper proposes pseudostationary networks at certain locations, specifically intersections on roads, which can be maintained and function without a central Roadside Unit to control vehicular nodes. Cellular networks facilitated communication despite user mobility thus the idea of using user mobility itself to facilitate communication may seem counter-intuitive at first but it forms the core of this paper. Keywords - VANET, Roadside Unit, Pseudo-stationary network, MATLAB

I.

INTRODUCTION

The idea proposed in this paper begins with the premise that VANET will be developed over few years to offer unique services needed for vehicular communication with significantly lower latencies than cellular networks and some existing services that are already provided by cellular networks at comparable latencies to that of cellular networks. The motivation behind the need to build new parallel VANET network when cellular network is already so widespread and easily available is beyond the scope of this paper[7]. However the need for government to exercise more control in communication area traditionally dominated by private carriers, as an independent backup s in case of emergency scenarios, clear advantages in terms of road safety, planned facilitation of services given the advent of automatic vehicles in near future could be cited as few of the reasons for promotion of VANETs. The higher latency of connection establishment in cellular network is a critical factor limiting the implementation of VANETs based on already existing and widespread cellular networks[2]. A car applying a sudden break on a rainy day would want to convey that information to the cars behind it as fast as possible. Use of existing cellular network to convey this information to cars behind would be always limited by the latency during connection establishment through an inherently centralized cellular network. VANETs would allow much lower latencies between direct or nearby nodes. Also, as cellular network is optimized around individual user and not users in a given location, VANETs propose lucrative solutions for certain time-critical services.

But VANETs can also be used to go beyond the time-critical vehicular signals exchange. Given the colossal number of vehicles on road, VANETs can be used for many more services from infotainment to mobile-commerce just like the cellular service however at a lower speed in initial phase. But when viewed as a supplementary service in this aspect to cellular one can allow to build richer set of applications[4]. Significance and eventuality of VANETs for short distance mobile nodal communication is sufficiently explored and proven and ranks higher than cellular on merit[4]. However, this paper aims solely at VANET as a network level service for larger distances which remains a place for discussion as cellular system already is very effective. Also, the need to build an infrastructure stacks against VANETs. This paper proposes a new ideas of pseudo-stationary network that can help reduce the outright need for building an allcomprehensive infrastructure of road-side units. There will remain the need to have some roadside units but the spatial density of them can be reduced by multifold with justifiable amount of delay in routing packets. Pseudo-stationary networks target at eliminating the need for roadside units every 1000m which falls in the range of 802.11p standard direct communication. If the RSUs can be spread sparsely, still allowing for the VANET system to function, scaling up would be more lucrative for entities investing in VANET infrastructure. For example, to determine the exact location of a user needs minimum 4 satellites in range. When the GPS satellite system was developed over a period of few years, system would not function in any area without the visibility of 4 nearby satellites. The system would be functional only when the complete infrastructure is in place. VANET systems faces similar problem which is exactly what this paper aims to address. During its development VANET system can still function, although at higher latencies even

without comprehensive network of RSUs thus making scaling up easier. This will also incentivize early adopters who can install modules in their cars & use it while the VANET network develops. Which in turn will provide valuable feedback from early adopters to the entities involved in system build-up. II. DESIGN AND SIMULATION

To test the feasibility of establishing PSN (Pseudo-Stationary Network) and its utility for routing network, a model was developed solely from scratch for this project. MATLAB was the only tool employed. The model allows the simulation of a small town with rectangular road network. Town is divided into 100+ blocks, each block containing 90-100 house-like entities. Roads constructed are bidirectional with single lanes with following variables 1. range of the Pseudo-Stationary Network on an approximate scaled based on the ratio of 802.11p range to that of average town block 2. directional probability distribution for flow of nodes 3. allowing for forming a chain among nodes flowing in the same direction - analogous to wagons of a train routing packets from engine to the last wagon in the chain Simulation generates interaction between any two nodes which can be categorized in three main types 1. interaction at an intersection- where there is a possibility of forming a PSN. If there are no stationary nodes when given node arrives at an intersection, the PSN is not formed. This indicates that when a vehicle arrives at an intersection, it can only avail the PSN based services if there are already some stationary nodes present at the intersection. If they are present, PSN can be established if it not already active.

2. interaction at a road stretch with a node travelling in opposite direction - a lot of useful information can be shared among the nodes travelling in opposite direction[6]. Node coming from destination of a given node can provide valuable information to any given node headed towards destination. Health of the traffic ahead, an accident warning or any other real time changes can be shared between the nodes travelling in opposite directions. The PSN system proposes another useful information if no PSN was found at previous intersection. The node travelling from A to C (after crossing B) can provide the information necessary for establishing a PSN at B to the node travelling from C-B-A. This can be used to form PSN even after the node going from A-B-C has passed B. 3. interaction between nodes travelling in same direction - this special arrangement of nodes can offer the highest speed for a message transfer as at any instance, the speed of message transfer when hopping from one node to another on a road stretch would be higher than the information carried by a node at any vehicular speed. Theoretically a standstill traffic on a given a stretch of the road will carry the information fastest across the endpoints than a node traveling on that very stretch which is completely empty at highest vehicular speeds. III. IMPLEMENTATION

Intersections are perfect match for this although slower traffic lanes, parking lots, public transport vehicles[3] could also be delegated as the spots for building pseudostationary networks. assumptions and approaches followed 1. The initial position of vehicular nodes is generated randomly unlike the real world scenario where there is a certainty in the direction flow of vehicles. Nodes in real world generally do not originate at random positions along the road stretches. This assumption however helps solve one significant problem as the random positions at which simulated nodes are generated can be viewed as part of a snapshot in random time sample of the continuous traffic. 2. The current number of nodes in the system are not tracked and only the initial number of seed nodes is controlled. This assumption helps get data required for this project in a better manner. To maintain a specific number of nodes circulating in the simulation would mean infinite number of node stacks as each node coming and leaving the simulation would need its own information stack. MATLAB simulation got significantly slower for nodes more than 2800 during simulation so static number of seed nodes were deployed and continuous inflow was discarded. However, outward flow of nodes is permitted. 3. The dimensions of simulation window were taken based an average town area spanning 100+ blocks with each block representing 10 rows and columns of houses like entities. The rule of thumb used was to scale the model based on the actual range of 802.11p (1000m) with respect to an actual city block size. 4. The directional flow of nodes is arbitrary with each node choosing 1 out of 4 directions at time of its generation using its index number as the random seed

The central idea is that the vehicles which have VANET modules installed in the cars, will collect the packets which could be live traffic information, emergency updates or network packets to sustain the network as a whole. These packets can be contained a geographically bound region without centralized monitoring of any roadside nodal unit. These geographical regions where pseudo-stationary networks will form in real time will be places of major nodal density.

Figure 1 Operation of PSN based information transfer

Figure 2 Flowchart for building & maintaining PSN

IV.

RESULTS
Figure 3 Nodes = 20; Range = 2 Figure 7 Nodes = 20; Range = 4

Figure 4 Nodes = 100; Range = 2

Figure 8 Nodes = 100; Range = 4

Figure 5 Nodes = 500; Range = 2

Figure 9 Nodes = 500; Range = 4

Figure 6 Nodes = 1000; Range = 2

Figure 10 Nodes = 1000; Range = 4

V.

CONCLUSION

VI.

REFERENCES

It can be concluded from the simulation results that with twofold increase in the nodes in the network could increase likelihood of forming and maintaining PSN or Pseudo-Stationary Networks by seven-eight times which aligns with the fact that any moving node can choose other three directions at a junction and the likelihood of interaction with a node in opposite direction would increase in the order of three for the entire network. With increase in the spatial range of nodes, the likelihood of building and maintaining the PSNs increased in the same order following a similar analogy. The delay involved in sending packets across network would be orders of magnitude higher than the cellular network (10s of seconds compared to ms). This however is the very area explored by this project and cost saved by not needing costly infrastructure to initiate the VANET based service can justify PSNs. The goal of this project is to address the initial chicken-n-egg problem while the VANET network is under development. It enables the functioning of the VANET based system without exhaustive network of Roadside units in place. There remains a need to have at least one RSU at certain points but the spatial density of the RSUs is no longer a prohibitive factor for functioning of the system It becomes a tradeoff against the speed of the message routing in the entire VANET.

[1] Sok-Ian Sou and Ozan K. Tonguz, Member, IEEE, "Enhancing VANET Connectivity Through Roadside Units on Highways", IEEE TRANSACTIONS ON VEHICULAR TECHNOLOGY, VOL. 60, NO. 8, OCTOBER2011 [2] M.J. Booysen, S. Zeadally, G.-J. van Rooyen, "Survey of media access control protocols for vehicular ad hocnetworks", IET Communications Received on 28th January 2011 [3] Tomoya Kitani et al. "Efcient VANETbased Trafc Information Sharing using Buses on Regular Routes", Vehicular Technology Conference, 2008. VTC Spring 2008. IEEE [4] Roger Michoud, Anna Maria Orozco, Gonzalo Llano, "Mobile Ad-Hoc Routing Protocols Survey for the Design of VANET applications", Intelligent Transportation Systems Symposium (CITSS), 2012 IEEE Colombian [5] Sean Hollister Engadget Editor, "NXP and Cohda teach cars to communicate with 802.11p, hopes to commercialize by 2014" [6] Jie Yang, Benyuan Lui, University of Massachusetts, "An intersection Collision Warning System using Wi-Fi smart phones in VANET", Global Telecommunications Conference (GLOBECOM 2011), 2011 IEEE [7] Abid, H. ; Tae Choong Chung ; Sungyoung Lee ; Qaisar, S. , " Performance Analysis of LTE Smartphones-Based Vehicle-toInfrastrcuture Communication", Ubiquitous Intelligence & Computing and 9th International Conference on Autonomic & Trusted Computing (UIC/ATC), 2012 9th International Conference

Вам также может понравиться