BLoc for Wireless Sensor Networks

BLoc is a statistical localization system that enables a wireless Sensor NETwork (SNET) to determine the physical location of its nodes. The coverage area is split into a number of user specified partitions and the system is able to identify the partition where any node resides. The only requirements for the system to operate are: (i) a sufficient set of the nodes are stationary and know their position, (ii) nodes can determine the received signal strength (RSSI) associated with any packet they can receive transmitted from any other node, and (iii) the system can use some set-up time to exchange "training" packets between stationary nodes and/or collect RSSI measurements at the stationary nodes from packets transmitted elsewhere in the coverage area. While operational, only a small subset of the stationary nodes need to be active and play the role of "clusterheads". Clusterheads can poll any node present in the coverage area and receive the packets this node transmits in response. Based on the RSSI measurements associated with these packets the clusterheads collaborate to collectively determine the transmitting node's location.

The methodology developed takes into account the fact that RSSI measurements are extremely variable, especially indoors, and can be very sensitive to even small changes in the position of the receiver or the transmitter. To that end, RSSI measurements are treated as being random quantities and there are no modeling assumptions been made on their attenuation as a function of distance. The underlying methodology and the system have been developed by researchers at the Boston University Center for Information and Systems Engineering led by Professor Ioannis (Yannis) Ch. Paschalidis.

A testbed of the localization system has been deployed in the the Manufacturing Engineering building at Boston University. The testbed uses MICAZ motes manufactured by Crossbow, Inc., which connect to a server by a set of Stargate motes. The testbed covers 16 rooms and several corridors by defining 60 partitions, covering a total of 5258 feet². It uses 12 clusterheads, that is, 1 clusterhead per 438 feet² on average. The testbed has achieved a mean error distance of D_e=8 feet, which is about the same as the mean radius of a partition. Compared to existing stochastic triangulation methods tested under the same conditions and on the same testbed, these results amount to a factor of 3.6 improvement in accuracy.

A different smaller-scale testbed with a much more dense set of partitions (36 inches² on average) has achieved a mean error distance as low as 9.26 inches.

Boston University Statistical Localization System (BLoc) for Wireless Sensor Networks
Home	Methodology	Map	BLoc	Contributors
BLoc is a statistical localization system that enables a wireless Sensor NETwork (SNET) to determine the physical location of its nodes. The coverage area is split into a number of user specified partitions and the system is able to identify the partition where any node resides. The only requirements for the system to operate are: (i) a sufficient set of the nodes are stationary and know their position, (ii) nodes can determine the received signal strength (RSSI) associated with any packet they can receive transmitted from any other node, and (iii) the system can use some set-up time to exchange "training" packets between stationary nodes and/or collect RSSI measurements at the stationary nodes from packets transmitted elsewhere in the coverage area. While operational, only a small subset of the stationary nodes need to be active and play the role of "clusterheads". Clusterheads can poll any node present in the coverage area and receive the packets this node transmits in response. Based on the RSSI measurements associated with these packets the clusterheads collaborate to collectively determine the transmitting node's location. The methodology developed takes into account the fact that RSSI measurements are extremely variable, especially indoors, and can be very sensitive to even small changes in the position of the receiver or the transmitter. To that end, RSSI measurements are treated as being random quantities and there are no modeling assumptions been made on their attenuation as a function of distance. The underlying methodology and the system have been developed by researchers at the Boston University Center for Information and Systems Engineering led by Professor Ioannis (Yannis) Ch. Paschalidis. A testbed of the localization system has been deployed in the the Manufacturing Engineering building at Boston University. The testbed uses MICAZ motes manufactured by Crossbow, Inc., which connect to a server by a set of Stargate motes. The testbed covers 16 rooms and several corridors by defining 60 partitions, covering a total of 5258 feet². It uses 12 clusterheads, that is, 1 clusterhead per 438 feet² on average. The testbed has achieved a mean error distance of D_e=8 feet, which is about the same as the mean radius of a partition. Compared to existing stochastic triangulation methods tested under the same conditions and on the same testbed, these results amount to a factor of 3.6 improvement in accuracy. A different smaller-scale testbed with a much more dense set of partitions (36 inches² on average) has achieved a mean error distance as low as 9.26 inches.