Low Energy And High Accuracy GPS Localization For Urban Environments

GPS localization has gradually penetrated into many applications in our daily lives,supporting a wealth of Location Based Services(LBS),such as transportation industries,outdoor navigation systems,wild animals tracking,and sports trajectory recordings.Currently,GPS based localization techniques have been widely used in vehicles,mobile phones and other terminal devices to provide easy-to-use location services.However,recent research work has shown that GPS suffers from two major problems.First,the internal GPS localization algorithm involves a great number of floating-point operations,with high computation complexity,which results in high power consumption.This imposes a serious constraint for continuous GPS navigation tasks.Second,GPS signals are affected by multi-path interference in complex environments such as urban canyons.It can lead to location error to tens or hundreds of meters,which fails to meet the application demands.To solve these problems,I have made the following technical contributions:· An Energy Model and the Low Power Algorithm:We first study the generic energy consumption mechanism of a typical GPS module.GPS energy factors are examined with black-box testing on various components,including the radio frequency,parallel computing units,CPU energy features,and key software procedures.Then,we present an universal GPS energy model,dominated by core computational units.The model covers the basic architecture of both scientific and commercial systems.To determine the optimal energy parameters,real GPS traces based benchmarks are implemented on an open source GPS receiver.With this energy model,we are able to determine the most power-hungry module,and present an energy efficient GPS localization algorithm with selective tracking.This algorithm chooses high-quality satellites with location contexts,to reduce the number of synchronous satellites as well as the overall local-ization power.Real road experiments show that GPS power consumption decreases more than 23.4%,with little accuracy reduction.· Energy Efficient Indoor/Outdoor Identification:Indoor/outdoor(IO)Identification provides very useful information for context sensing in mobile devices.We present a novel IO Identification solution,namely Sat Probe,which explores relationship between number of visible satellites and the IO status.Sat Probe determines the number of visible satellites with a lightweight existence search,instead of the GPS location calculation process with high computation complexity.Compared with the literatures,experimental results show that Sat Probe produces better I/O recognition accuracy,along with more than an order of magnitude performance improvement in both energy efficiency and detection time.· High Accuracy GPS Localization in urban canyons:GPS localization accuracy suffers from urban canyons.We discuss how to mitigate the urban canyon problem,with cooperative bike localization in a bike sharing system.Specifically,a group of sharing bikes are organized into a network,namely Bike Net.Volunteer users,who pass by these bikes,can open radio scanners,accelerometers and magnetic sensors,to measurement the distances and directions among these bikes,which generate the geometric constraint for Bike Net.Then,Bike Net maps all nodes' satellite range measurements to a single lead node's view.By selecting a high quality and geometrically diverse set of range measurements,the lead node can accurately derive the locations of the entire network.Bike Net is evaluated in both residential area and commercial area.Compared with traditional GPS method and node level cooperative methods,Bike Net produces a significant improvement in localization performances,solution availabilities,and robustness.