Practical Lessons from

I n pervasive computing devices, a device’s ability to discover its location often proves useful. To achieve this, many research and commercial location sys-tems have used technologies such as radio signal strength, radio time-of-fl ight, ultra-sonic time-of-fl ight, infrared proximity, and optical vision. The most successful system, the Global Positioning System, is satellite-based, so it works worldwide. Inexpensive GPS receiver chips are also easy to build into devices. How-ever, because GPS receivers require a clear view of the sky, they typically don’t work indoors, under cover, or in many urban canyons. When we started our work in 2003, most location systems other than GPS relied on infrastruc-ture that limited coverage to a room or build-ing. Inexpensive ubiquitous positioning systems didn’t exist, slowing the widespread adoption of location-enhanced computing. We built Place Lab to address the need for a ubiquitous loca-tion capability. The project Place Lab lets a device estimate its location by scanning for fi xed radio beacons—such as nearby 802.11 access points and GSM (Global System for Mobile Communications) cell towers—and referencing the beacons ’ positions cached in the mobile device’s database. To enable real-world deployment, we set these goals for Place Lab: Work over a wide area, indoors and out. We chose 802.11 and GSM signals, which are being rapidly and ubiquitously deployed throughout populated areas. Run on commodity devices. Place Lab uses everyday device hardware and whatever radios a device is equipped with. It can run on laptops, PDAs, and cell phones with dif-ferent architectures and operating systems. Observe privacy needs. Users ’ devices esti-mate their locations locally, without requir-ing a network connection or any server-based infrastructure. Support standard programming interfaces. To support existing applications, Place Lab exposes its location estimates in industry-standard APIs, including NMEA 0183 (th