We propose an ad-hoc peer-to-peer network of vehicles for traffic management and control. Each vehicle is a node in the system. If any particular user requires traffic information (not just about the road that she is driving on, but other roads) she can log onto this network by plugging his PDA into the sensor node socket. Therefore, a query can be generated by individual sensor nodes. However, this is not required. Even if the driver is not interested in getting additional information, her car acts like a network node that collects, aggregates and forwards information.

This system is quite concurrent to the crude manual technique currently used to collect pertinent information, such as asking another person coming from that region about the possible condition of that area. “Hey Robin, how is the traffic condition on I-95 S, 30 miles from this place” – the answer can be “ ahh , there is traffic jam out there “ – another answer can be “ I am not sure, I just got onto I-95 few minutes back, but I did see some cars coming from far off I-95 S, hence the condition SHOULD be good”.  Here, sensors may replace this task. 

1.2. Current Work

Our idea is antithetical to much of the current work in this field. Current research at UC Berkeley [1], [2], UCLA [3] and MIT [4] share the same architecture – they are relying on sensors that are located on the roads/highways.

Work at UC Berkeley for development of Fjording [2], an architecture for queries over streaming sensor data, uses sensors as a test bed for traffic management.  Envisioned is a traffic management system in which miniature sensors are scattered on a freeway. Battery driven, minute sensors powered with a RF Transmitter will be able to gather information like velocity and location of the vehicles passing over them. These will transmit vehicle information to the proxy / base station which can aggregate/ process the information and answer queries.

WINS (Wireless Integrated Network Sensors) [3] at UCLA combines sensing, signal processing, decision capability, and wireless networking capabilities in a compact, low power system. WINS can be used to form a network of sensors providing multi-hop communication. Micro-power WINS systems-on-chip can be used for traffic management by deploying them on the highways/freeways for traffic monitoring. 

LEACH (Low Energy adaptive Clustering Hierarchy) [4] from MIT is designed for applications where end-user wants to remotely monitor the environment. LEACH makes use of distributed cluster formation, local processing to reduce global communication, and randomized rotation of the cluster-heads. Being energy efficient, the LEACH protocol architecture provides a good solution for traffic management.

Forming a sensor network on the highway/freeway and aggregating data to gather traffic information have the following limitations:

James A. Misener, et al., [5] describes various tools for a vehicle-vehicle and vehicle-highway sensor topology. They are focusing on applications like cooperative collision avoidance system where the cooperation is limited to few vehicles at a time. They studied various sensing system including passive electro-optical sensors, active electro-optical systems and radars in context of "Sensor Friendly Highway". Their vision towards a sensor-friendly highway is restricted towards communication between two vehicles and has no notation of a network of vehicles collaborating to give results. Still, we can use their study to infer that vehicle-vehicle collaboration using sensors seems like a good approach, although not trivial to implement. They also produced results that supported the use of Radar (RF based) for detecting rows of cars, tracking of another vehicle, even with interfering vehicle maneuvers, a pick-up truck that cuts in between vehicles etc. This greatly helps us in moving forward with our design.

Advanced Transportation Management Systems ([1],[6]) covers various areas of Traffic Detection, Traffic management, Traffic Theory and modeling and Decision Support. In the Data Collection arena, PATH is using 12 fixed video cameras and 2 pan cameras, which provide continuous view of I-80 between the Bay Bridge approach and University Avenue. A step towards building Intelligent Transport System (ITS) used in ATMIS is to utilize the vehicle waveforms obtained using inductive loop detectors (ILDs). These "signatures" can be re-identified at the downstream stations, and yield statistics like real-time section travel times, section speed, and section density. The Video-base Signature Analysis and Tracking (V2SAT) System is begin used to non-intrusively track individual vehicles on freeways for data collection. It involves the use of computer vision methods to make optical measurements on digitized real-time images of each vehicle. For each vehicle, a numeric Video Signature Vector (VSV) is generated and transmitted by the detection module to a centralized computer. Theses VSVs are used to re-identify vehicles. Real-Time Laser-based Detection System for Measurement of True Travel Time on the Highway [7] is used for real-time measurement of moving vehicles.  The primary goal is to develop a roadway detection system that can be used to gather reliable travel time data non-intrusively.

These methods have the inescapable limitation that they require a considerable infrastructure to be installed on the highway. Moreover, maintenance becomes a big issue. Although this may not be a problem in some parts of the worlds, a system which have a high infrastructure cost may not be a solution for most countries. Furthermore, visual sensors and video based solution cannot cope with very high traffic situations where there is high density of traffic that needs to be tracked. 

Jean-Luc Ygnace, et al [8] describe a method to use cellular telephones to gather travel time information. This approach is orthogonal to the estimation techniques based on data from road sensors embedded in the pavement. Cellular telephone positioning does provide a cost-effective method for tracking a vehicle and inferring conclusions like travel time, and location. The report states that with 5% of freeways travelers equipped with a live cell phone, one can predict a 95% accuracy in freeway link travel time estimates. 

The Federal Communication Commission Phase II mandates for Enhanced-911 [9],(E-911) requires that wireless carriers must provide the location of a 911 wireless call to the Public Safety Answering Point (PSAP), within approximately 125 meters. Mark Birchler, et al, [10] describes a overview the technical details of how such a system is made possible.

Cellular positioning paves the way for the cost effective solution for traffic management. However, the basic problem in this approach is that of security and privacy. Another fact that remains is that the cellular phone network is not available all around the globe and the concentration of cellular phone penetration varies tremendously. E-911 is applicable in USA, where also many of the cellular phone providers are not able to provide the same. Asia and Europe have no such laws and it will be difficult task to convince the Cellular Service Providers to give this service.

This brings us to the conclusion that we need to concentrate on a service which makes least assumptions about the network (both social and data) around. Ad-hoc networks are an ideal solution to such problems. Unfortunately, large-scale deployment of sensors on the highway is not a solution for global acceptance of the system.