Introduction

Location aware computing provides applications with knowledge of the physical location where the computation is taking place. This allows applications to operate in a more context-sensitive fashion. Fundamental to location aware computing is the task of determining the location of the computational device. In outdoor environments with unobstructed views, GPS is widely used for this purpose. Indoors, and in crowded city streets, however, the effectiveness of GPS is greatly diminished. A number of approaches have been made towards indoor localization, with varying features and measures of success.

We present a system that provides an infrastructure for location aware computing that emphasizes the following key features.

• Cost The system must not be prohibitively expensive, and its costs should scale well with its size.
• Deployment Our system relies on technology that is already widely available and in use today. The hardware is multi-purpose and can be used for a variety of other computing tasks when not being used for localization purposes. Many potential users of our system would not need a significant investment in capital or other resources to take advantage of our infrastructure. Additionally, the system is simple and almost trivial to deploy on a large scale.
• Privacy Previous studies find that privacy is a significant concern in ubiquitous computing. A location aware system should preserve the privacy of its users. Users should be able to choose whether or not to reveal their location and identity to others, and should not be actively tracked without explicit consent. It should not be possible for the system to track users without permission.

Approach

Our infrastructure consists of a large number of statically positioned bluetooth devices that are set to be globally discoverable (i.e. they will respond to bluetooth device inquiries). The locations of these devices, which we will refer to as beacons, are fixed and entered into a database. We use low-cost USB Bluetooth adapters in our system, but any bluetooth device that does not change position can be used as a beacon in our infrastructure.

A bluetooth-enabled device (e.g. a cell phone, pda, or laptop) that wishes to determine its position can execute a simple algorithm

1. Scan for nearby beacons by performing a bluetooth device inquiry. This involves repeatedly broadcasting a predefined sequence of bits while hopping frequencies pseudorandomly. The broadcast contains no content which can be used to identify the locator. Beacons that detect the broadcast will reply with their own identities.
2. The locator determines the location of detected beacons. A cache is maintained, mapping beacon identities to their locations. If the location of a beacon is not present in the cache, then the locator can query the beacon directly at the cost of anonymity.
3. The locator determines its location relative to the detected beacons. If only one beacon is detected, the locator can conclude it is within 10 meters of the beacon. With multiple beacons, simple geometry can be used to refine the estimate.

The key to creating a trivially deployed infrastructure lies in the complexity and cost of the beacons. Previous systems have all been prohibitiely expensive to deploy on a wide scale or required specialized hardware. Our beacons are composed of two off-the-shelf components which can be purchased for a total of US$30. The first is a USB Bluetooth adapter that would typically be used in a desktop or laptop PC. The second is a USB hub with an AC/DC adapter. These parts can currently be purchased retail for $20 and $10, respectively. Once the bluetooth adapter is attached to the USB hub and the hub is powered on, the beacon is initialized by connecting the hub to an open USB port on a laptop computer. The laptop sets the bluetooth adapter to discoverable mode and can then be disconnected. The beacon remains discoverable by other bluetooth devices until the USB hub loses power. Naturally, this raises the concern that the entire system will need to be re-initialized if, for example, the building loses power. We expect this to happen infrequently, however, as modern industrialized buildings rarely lose power.

By deploying beacons in this manner, we require no infrastructure beyond electrical power. The exact placement of a beacon is flexible and can vary with the design of the building and the desired positioning precision.

Progress

We are in the process of upgrading our previous bluetooth infrastructure, which was similarly constructed, but relied on PCs to host bluetooth beacons. This previous scheme was much more prone to failure and corruption, as computers were often decommisioned or physically moved. We have built software clients to enable java cell phones, Linux PDAs, and Linux laptops to leverage our infrastructure and provide applications with location awareness. We have investigated methods to improve the accuracy with which we can estimate the position of a device using the bluetooth infrastructure. We have used signal strength measurements of inquiry responses combined with probabilistic inference techniques to refine the position estimates.

Future

In the near future, we plan on fully deploying this infrastructure throughout the Stata Center at MIT. We are also continuing our work on improving the accuracy of the system by systematically decreasing the radio range of the beacons in select areas.

References

[1] G. Anastasi and R. Bandelloni and M. Conti and F Demastro and E. Gregori and G. Mainetto. Experimenting an Indoor Bluetooth-Based Positioning Service. In Proceedings of the International Conference on Distributed Computing Systems Workshops, pp. 480--483, May 2003.

[2] Louise Barkhuus and Anind Dey. Location-Based Services for Mobile Telephony: a study of users' privacy concerns. In Proceedings of Interact 2003, pp. 709-712, Zurich, Switzerland, 2003.

[3] Bluetooth Special Interest Group. Bluetooth Profile, Specification of the Bluetooth System, Version 1.2, Nov. 2003.

[4] A. Harter and A. Hopper and P. Steggles and A. Ward and P. Webster. The Anatomy of a Context-Aware Application. In Wireless Networking, pp. 187--197, vol. 8 num. 2/3, 2002.

[5] M. Nilsson and J. Halberg. Positioning with Bluetooth, IrDA, and RFID.

[6] N. Priyantha and A. Chakraborty and H. Balakrishnan. The Cricket Location-support system. In Proceedings of the 6th annual international conference on Mobile computing and networking, pp. 32--43, 2000

[7] Author A. Lastname and Author B. Lastname. Title of the paper. In The Proceedings of the Conference, pp. 1-3, City, State, Country, Month 2004.

[8] I. Getting. The Global Positioning System. In IEEE Spectrum pp. 36-47, vol. 30 num. 12, December, 1993.