Loca­tion is a fun­da­men­tal ser­vice for mobile com­put­ing. Typ­i­cal GPS receivers, although wide­ly avail­able, con­sume too much ener­gy to be use­ful for many appli­ca­tions. Observ­ing that in many sens­ing sce­nar­ios, the loca­tion infor­ma­tion can be post-processed when the data is uploaded to a serv­er, we design a Cloud-Offloaded GPS (CO-GPS) solu­tion that allows a sens­ing device to aggres­sive­ly duty-cycle its GPS receiv­er and log just enough raw GPS sig­nal for post­pro­cess­ing. Lever­ag­ing pub­licly avail­able infor­ma­tion such as GNSS satel­lite ephemeris and an Earth ele­va­tion data­base, a cloud ser­vice can derive good qual­i­ty GPS loca­tions from a few mil­lisec­onds of raw data. Using our design of a portable sens­ing device plat­form called CLEO, we eval­u­ate the accu­ra­cy and effi­cien­cy of the solu­tion. Com­pared to more than 30 sec­onds of heavy sig­nal pro­cess­ing on stand­alone GPS receivers, we can achieve three orders of mag­ni­tude low­er ener­gy con­sump­tion per loca­tion tagging.

• RESEARCH PUBLICATION