Diplomarbeit
Optimization of Data Dissemination and Personalized Queries on Information in
Vehicular Ad-Hoc Networks
Completion
2006/08
Students
Daniel Graupner
student
Advisers
Description (German)
General Overview
In recent years, the application of mobile ad-hoc networking among vehicles has become an
important research field for the automotive industry. Vehicular Ad-Hoc Networks (VANETs)
consist of vehicles that are equipped with wireless communication devices based on the
standard IEEE 802.11 Wireless-Lan. Embedding mobile ad-hoc network devices into vehicles
will pave the way for future Advanced Driver Assistance Systems. Car-to-car communication
could be used to improve road safety and comfort for the driver as well as the passengers.
Consider the situation in which you drive into an unfamiliar city. Your vehicle navigation
system checks the traffic conditions and requests parking information in the vicinity of
your destination on-demand and directs you quickly through the city, avoiding busy roads
and road blocks, to a free parking space.
Network-wide personalized queries
Personalized search for information is another valuable application that can improve
passenger's comfort. These network-wide personalized queries are a specific research field
at Volkswagen. The current situation reflecting the functionality and the basis for this
diploma thesis is described as follows:
* several vehicles (equipped with sensors) form an IEEE 802.11 ad-hoc network in a city
centre (concrete street model of Braunschweig is available)
* every node broadcasts its complete information (from its environment) in regular
intervals to every node in its communication range
* received/detected information is stored in a cache (knowledge base, KB, cache) and is
replaced if older information already exists
o the cache-size is limited to 20 entries corresponding to 13 parking places, 5 gas
stations and 2 restaurants
* vehicles towards the city centre request information about their location, e.g. ask for
free parking spaces or traffic conditions on their way
o query reaches target area through locally optimized (greedy) position-based routing
(reactively
o cars (nodes) that have the required information respond to the query
The given search strategy is not optimal because it does not take the importance of
information into account and thus wastes bandwidth by transmitting outdated information.
The perfect scenario
In a perfect scenario every vehicle has a complete and up-to-date view on its environment
and queries on information will always be (correctly) answered in time. Due to the nature
of mobile ad-hoc networks the perfect scenario differs from real world scenarios
completely. Vehicles move rapidly and unpredictably from one network partition to another.
Therefore requests may get lost and caches become invalid. Also in big cities several
thousands of vehicles may take part in a VANET. During rush hour it is most likely that
all of them enquire for information. Although bandwidth is limited every query should be
answered.
The problem to solve and methods of resolution
Information in the KB will not be current after a certain period of time. This fact
crucially depends on the type of information, e.g. free parking place information becomes
obsolete faster than locations and fuel prices of gas stations. Therefore a function
representing the relevance of the information has to be developed. With this function the
cache size can be optimized and the frequency can be reduced. It is also important to
examine several query strategies. According to a very dynamic topology (unpredictably
changing network partitions) the answer should be provided as accurately as possible. One
can think of different approaches to solve this, e.g.:
* if the query passes a node that is able to satisfy the return condition (answer the
query), the query stops and the answer is routed back to the sender
* if the query reaches the target area, the location is flooded (selective flooding,
gossiping) and every node holding the requested information routes back the answer to the
sender. The sender analyzes the data and incorporates the information in its cache
* an intermediate node bundles the results in the target area, analyzes the data and
accomodates demand
* also hybrid approaches may be reasonable
The?following subtasks must be examined:
* research and evaluation of basic approaches to find a possible solution for the
described problem. Finding a relevancy-function that fulfills the established criteria
* hight scalability is desired, i.e. answering thousands of queries
* minimize network load by more intelligent, content-aware broadcasting
* simulate the most promising methods of resolution in order to compare them on an basis
of defined conditions
