University of Minnesota 1 / 9 May 2011 Energy-Efficient Location-based Services Mohamed F. Mokbel Department of Computer Science and Engineering University of Minnesota
May / 9 Location-based Services: Now Location-based traffic reports Range query: How many cars in a freeway? Shortest path query: What is the shortest path to my destination? ■ Location-based store finder Nearest-neighbor query: Where is my nearest restaurant? Range query: What are the restaurants within one mile from my location? ■ Location-based advertisement Range query: Send e-coupons to all customers within five miles from my store
May / 9 Future
May / 9 Energy Consumption in Location-based Services Power consumed by location- detection devices It is crucial to minimize the power consumption of such devices to keep them alive longer ■ Power consumed at the database server to answer location-based queries Most of location-based queries are inherently continuous, which makes them expensive to evaluate on the server side
May / 9 Energy Consumption in Transportation Source: EIA. Annual Energy Review Table 2.1a Energy Consumption by Sector, The transportation sector consumes 29% of the US power Source: BTS, National Transportation Statistics. Energy Consumption by Transportation Mode in the United States, (in Trillion BTUs) Road accounts for about 80% of all the energy consumed by transportation in the United States and this share has remained constant in time.
May / 9 Energy Saving in Mobile Devices Most of the energy consumption in mobile devices is consumed in detecting/uploading the user location Approaches of energy saving Sampling. Update the location information every t time units Prediction. Send the predicted future trajectory, then, send an update only if different from the predicted trajectory Need to go beyond data-driven techniques to query-driven techniques where the location will be uploaded only if it will affect the result of a given query
May / 9 Energy Saving at the Server Side ■ Minimize the work that the DBMS needs to do through a built-in structure ■ Power-Aware Evaluation of Continuous Queries Two approaches for continuous query evaluation: A set of consecutive snapshot queries, executed every t time units Incremental evaluation Power-aware Cost models for incremental evaluation and shared execution Shared execution Load Shedding DBMS GIS LBS Layered Approach GIS Interface LBS-Index LBS Query Processing DBMS Built-in Approach
May / 9 Energy Saving in Transportation Personalization: Giving the right answer is essential in saving driving time We need to go beyond the traditional nearest-neighbor queries that are solely based on distance to consider more context and preference-aware queries Accurate traffic prediction Prediction and avoidance of traffic congestions save driving time We need to devise “long-term” and “accurate” prediction techniques that send alerts about possible congestions Shortest path queries Finding the right shortest path route significantly affect driving time We need to go beyond the typical shortest path algorithms that mostly consider the distance to consider the time of the day, and time-aggregated graphs
May / 9 Thanks