My principal interest in mobile applications is to push the boundaries of innovation to create uniquely mobile experiences. I hope my blogs excite and challenge developers to think 'outside the box'.
searching location based services on your phone
coultonp | 04 July, 2006 13:35
One of my PhD students Paul Gilberstson is working on methods related to algorithms that will allow searches of location based information on your mobile phone rather than through a server (although periodic data refreshing is required from the location server). The advantage being that information can be processed and presented quickly thus avoiding network latency issues.
The algorithm will be presented in detail at PIMRC 2006 in
The data is organised and stored into a series of five circular containers, or what we have termed ‘buckets’, based on distance from the initial position. Each bucket contains information about the minimum and maximum distances of data that it contains, and the location of the centre point for this circular area. Initially the complete online positional database is queried and data is organised into the buckets. Data points that fall outside the fifth bucket are ignored.
When a user’s position changes, the data within the inner most bucket is checked for positional collision. Since only the innermost bucket of data is checked the computational requirements on the phone are kept to a minimum. If the users position changes such that data in the inner bucket is no longer valid (in other words the users position has moved towards the edge of the bucket) the data is invalidated. The system is realigned so that the users position is once again central to the system and data that no longer needs to be checked moves outwards, and new data moves into the inner bucket and is thus can be checked for collisions. When the outermost bucket is invalidated, the server side database is queried for data points which now fall within the outer bucket.
Paul’s results show that the algorithm presents a very significant improvement over brute force methods, for example, with a database of 100,000 data points a 99% reduction in processing time is observed.
Having mobile side searching offers some interesting possibilities and for those of you attending PIMRC I am sure Paul will happily provide a more complete explanation.
Location Based Services |
Next | 
Previous |
Comments (0) | 
Trackbacks (0)
Comments
You must login to post comments. LoginWho am I?
My principal interest in mobile applications is to push the boundaries of innovation to create uniquely mobile experiences. I hope my blogs excite and challenge developers to think 'outside the box'.
Calendar
Recently...
Categories
