New Penn State System Improves Wireless Access To Global Information Sharing

Dr. Ali Hurson, professor of computer science and engineering, says, "A solution does not currently exist that provides an effective means to cope with all the issues involved while accessing a massively diverse amount of data over a wireless connection. This is particularly true for devices with limited processing capacity and resources such as cell phones, palm pilots, pagers or other hand held devices."

Hurson and James B. Lim, who earned his doctorate at Penn State in 1998, have, however, proposed a new concept they call a mobile data access system (MDAS) that can tap heterogeneous data sources -- stock quote services, news, airline information, weather, the Internet or the World Wide Web -- through both wired and wireless connections.

MDAS works by superimposing a multidatabase system on a wireless-mobile environment. Multidatabases give users integrated access to multiple databases with a single query. They appear to the user to behave like a super-search engine that not only locates information from multiple sources but also combines and integrates it. For example, a single query to a multidatabase could not only produce a list of bed and breakfasts in State College, Pa., but might also tell you their rankings in two different on-line travel guides -- a task that would require accessing at least three databases and integrating the resulting information.

MDAS also provides an efficient way to manage the query traffic so that gridlock does not occur when more than one user wants to access the same database.

"Current control schemes within existing wireless mobile computing environments do not efficiently manage these accesses because they do not address the limited bandwidth and frequent disconnection associated with wireless networks," Hurson notes. "Our approach includes a concurrency control algorithm, called v-lock, which uses global locking tables to serialize conflicting transactions and detect and remove deadlocks."

In simulations run at Penn State, V-lock was as fast as its nearest competitor, but allowed about 20 percent more transactions at the highest levels of conflict situations.

Hurson explains that the "easy way" to prevent two users from creating gridlock is to stop one transaction. V-lock, however, allows both users simultaneous access to the same database, as long as they want to go to process different parts of the information source. For example, two users who want to process information from the same weather database would be allowed access simultaneously if they were searching for forecasts for two different cities.

When the query comes in, V-lock has to "guess" whether or not the two users want access to the same parts of the database. Semantic information contained within the locking tables enables v-lock to examine the query and make an "educated guess" when directing traffic, Hurson says.

Providing the user with information "anytime, anywhere" is the ultimate goal, he notes. Someday, while waiting for a taxi, you may be able to pull out your personal digital assistant, tell it to check your calendar and the weather reports, and, if it's not going to rain during your next free weekend, book you a stay at a top-rated bed and breakfast in State College.

But we're not there yet.

Hurson says, "Although the results we have demonstrated are very promising, the work still needs to be extended in a variety of ways. However, users are demanding new more intelligent services and we're going to have to meet these challenges."

MDAS is described in Heterogeneous Data Access in Mobile Environment - Issues and Solutions, a chapter in Advances in Computers, Vol. 48, released Aug. 31-Sept 3. The research was supported, in part, by grants from Lockheed Martin and the Department of Defense.