Microsoft Research is working on a GPS technology that gets around high power requirements by offloading the heavy number crunching to the cloud. MIT Technology Review reports that the company is developing something called Cloud-Offloaded GPS (CO-GPS) and is testing a new mobile sensing platform called CLEO that puts the idea into practice. The research group’s findings were recently published in the 10th ACM Conference on Embedded Networked Sensor Systems. Right now, the GPS antenna in your smartphone is its biggest energy hog, completely running down the battery in about six hours if it’s left on constantly. The research team shows that using CO-GPS they can theoretically get constant sensing for a year and a half with one measurement per second granularity using just two AA batteries.

Ordinarily, in order for your phone to get a fix on your location, it needs to work for 30 seconds or so to pull down the necessary information from orbiting satellites. There’s also a lot of heavy signal processing work that needs to be done just to acquire a connection and track the satellites as they move. Researchers are finding that by tweaking things just so, and offloading the heavy lifting to the cloud, they can get an initial fix from just a few miliseconds’ worth of data. The cloud service also leverages publicly available information like satellite trajectories and an Earth elevation database to do more with less, getting power consumption that’s orders of magnitude lower than what we currently have. For example, to get an initial GPS fix on a smartphone with current technology takes about one Joule of energy. Using the researchers’ technique cuts that to 0.4 millijoules — a 99.96 percent cut.

The team believes that the dramatic improvement in efficiency will lead to new services based on continuous GPS logging; for example, building a database of noise pollution levels in a given city, or getting tailored directions or search results based on the routes you most commonly drive. The team also hopes to push the efficiency of its solution even higher by experimenting with new compression techniques and refining its signal processing algorithms.