The RFID RSSI measurements taken with Queralt on April 7, 2008 can be found in the attachment.

Measurement data has been analyzed.

The result is uploaded with MATLAB script file.

Measurement result

1. Average RSSI data over distance for all RFIDs

2. The average RSSI over distance for all RFIDs

2. The RSSI variation over distance for different RFID orientations

Vertical

Horizontal

Random

For all RFIDs

3. Distance data over Measured for all RFIDs

Horizontal

Vertical

Random

All RFIDs

Conclusion

The measurement data statistics shows two major observation.

First, the estmation of RSSI given distance is quite doable. Speically, we collect many data, and average out, then the average RSSI over disatnce shows strong linear correlation over distance.

Second, the estimation of distance given RSSI might not be possible. Given one measurement of RSSI, the absolute variation of distance is often more than 100 %. which means that the estimation error from a RSSI value could be more than 100 feet. The biggest challenge in this problem is the minimum possible distance given RSSI is unpredictalbe. That means even a large RSSI value could come from very short distance between reciever and transmitter. Therefore, it might not be good idea to use direct measurement of RSSI for estimating distance. However, it might be worthwhile to looking into possiblity to use joint statistics of different set of RFIDs to enhance accuracy of estimation.