Real Time Traffic Information Dissemination

Revised July 25, 2003

There are generally two different schemes for providing real time traffic information for telematics and navigation system applications.  The first, and most developed, is a network of fixed infrastructure sensors, typically at intersections and over busy routes.  Other than for Intelligent Highway Systems (IHS) and traffic control device use, the direct use of these sensors is quite limited within vehicles.  The second method involves in-vehicle sensors.  The most critical information for telematics and navigation system use is traffic flow, and this discussion will be limited to this aspect, bearing in mind that other information may be provided within an information dissemination infrastructure.

One method for detecting traffic is to monitor the movement of radio transceivers, such as cellular telephones, with respect to fixed infrastructure.  Such systems can passively track vehicle movement, distinguishing extravehicular and parked vehicles from vehicles moving with the flow of traffic on a predetermined route.  A central database produces a map of traffic flow patterns on the predetermined routes, which may then be broadcast or communicated to the vehicles, for example using the very same transceivers employed for tracking.  This method leverages the existing cellular infrastructure and high market penetration of cellular telephones in vehicles, as well as the growing intelligence of cellular handsets to be able to provide a user interface for a sophisticated navigational system.

Alternately or as a supplement to a network-based traffic monitoring system, it is readily possible to use sensors within the vehicle, which may be more specific or "richer" in information content than simple tracking data.  These sensors include vetronics (vehicular electronics bus) data, and aftermarket sensors, either of which may include GPS.  Of particular interest is the video camera, since this, in conjunction with intelligent analysis (either by machine or human), can lead to causal or analytical information relating to traffic patterns.  Thus, an image of a pothole may be more useful in modifying traffic patterns than simple traffic congestion information.

The present business models for cellular carriers do not allow for cost-efficient transmission of unsolicited video images, as would occur in an automated telematics system.  Therefore, I propose the use of unlicensed spectrum for broadcast of such images.  In particular, variants of 802.11 and other wireless data networks provide advantageous short range, high data rate capabilities.  Organization of multiple nodes into an ad hoc multi-hop and/or store-and-forward network allows broad dissemination of the data, meeting "real time" criteria, especially where node density is moderate to high.  It is important to understand that, in navigational applications, real time latencies can extend from seconds to minutes.  Therefore, the network control paradigm can prioritize data transmissions and employ multicast protocols, conserving bandwidth.  Likewise, each node may act as a filter and intelligent processor to truncate redundant or superfluous communications.

A proposed implementation of the vehicular sensor network includes a forward-looking tilt/pan/zoom digital camera, preferably inside of the windshield, having sufficient frame rate to preliminarily identify events of interest, and sufficient resolution to capture a quality image of the event.  These may represent two different modes of operation, or even different imagers.  The image information is processed, in conjunction with precise location information, and stored in a local database.  This local database may be purged as fresh or "better" information regarding the same event is received, either directly or through a communications link, or as it expires.  A transceiver is provided to link the vehicle with neighboring vehicles.  For example, 802.11g may be suitable.  A high gain or "smart" antenna, e.g., mechanically or electronically steerable, may be used to improve performance.  As vehicles come within communication range, they negotiate a communication, which may be point-to-point or multicast, and transmit information regarding stored events.  The events may be prioritized, for example according to an inferred real time latency criterion.

Typically, the communication will include various metadata regarding the event, such as location, time, and inferred classification, and may also include a communication of the raw data, such as vetronics bus status and compressed image.  This later information may be quite useful to allow a driver to understand the event, in a manner which automated analysis would or could not emulate.  Further, such information may be efficiently presented and quickly processed by the driver with minimal distraction.

The user interface of the device for the driver will typically include a voice navigation interface and graphic display.  A graphic user interface, with pointing device built into the steering wheel may also be provided.  Events are sorted on a navigation priority basis, that is, in a manner relevant to the driver.  Thus, while the local database may include information which is quite irrelevant to the driver, supporting the network operation of the system, the user interface is filtered for relevance.  The driver's route may be pre-entered at the beginning of a trip, or inferred from a map database, collaborative filters, past history, interactive use, or the like.

By leveraging mass produced consumer technologies, such as digital cameras, wireless LANs, GPS, and PDAs as the core elements of the system, costs may be kept low, and the components may be used for other purposes, inside and outside of the vehicle.