Understanding the characteristics of low-power wireless links and radios is an essential step towards building robust, efficient and reliable wireless sensor networks. In this project we study and evaluate the fidelity of the Received Signal Strength Indication (RSSI), which the low-power radios use to measure the power of the wireless signal. This value is heavily utilized in many wireless sensor network protocols and applications, such as localization, topology control, link scheduling, and link quality estimation. With extensive experiments, we show that inaccuracies in the RSSI values reported by widely used 802.15.4 radios, such as the CC2420 and the AT86RF230, have profound impact on these protocols and applications. Therefore, we also developed a calibration scheme to effectively minize the adverse effects associated with inaccurate RSSI values.
Stepping up to the link layer, we note that packet loss and energy consumption in sensor networks depend critically on the quality of the network's wireless links. Experimental results have shown that a low-power wireless link can be in one of three states or 'regions', as the inter-node distance increases: connected, transitional (gray), and disconnected. Moreover, the transitional region spans a sigficant portion and is likely to be even larger than the connected region. Therefore, in this project we explore the characteristics of the transitional region and study the possibility of picking reliable links within this region.
Calibrating RSSI Values Reported by 802.15.4 Radios
IEEE 802.15.4 standard specifies that a radio's PHY layer must provide an 8-bit integer value as an estimate of the received signal power. This value is commonly known as the Received Signal Strength Indication (RSSI) in the wireless sensor networks (WSN) community. Numerous WSN protocols use RSSI measurements extensively, including those for localization, link quality estimation, packet reception ratio modeling and transmission power control. While many protocols directly use the RSSI measurements that the radios provide, the standard only requires that the reported RSSI values should be linear and within ±6 dB of the actual RSSI values. However, ±6 dB is a wide error margin. For example, Packet Reception Ratio (PRR) can decrease from 100% to 0% with a 2 or 3 dB difference in the received signal strength. The consequence of this observation is that possible inaccuracies in the reported RSSI values can profoundly impact applications that rely on RSSI measurements.
In this project we examine two 802.15.4 compliant radios, the widely used Chipcon/TI CC2420 and Atmel AT86RF230, and show that they do indeed introduce systematic errors in the RSSI measurements that they provide. As a matter of fact, the coarse RSSI value vs. input power graph included in the CC2420 datasheet hints at the existence of non-linearities. Nevertheless, the manufacturer states that the RSSI response curve is very linear. We independently derive high resolution RSSI response curves using a variable signal generator and verify the existence of the non-linearities hinted by the CC2420 datasheet (we also note that the AT86RF230 datasheet does not provide an equivalent graph). Fortunately, these response curves are radio-specific but device independent. In other words, different physical devices that use the same model of radio have identical response curves. Consequently, mitigating these nonlinearities does not require calibrating each device individually. This result allows us to develop a generic calibration scheme to compensate for the radio's inaccuracies.
The data and instructions on calibrating RSSI measurements provided by the CC2420 radio chip are available on this page.This work appeared in the proceedings of EWSN 2010: Download PDF and Slides
| Spatial Characteristics of the Gray Region for 802.15.4 Radios
Packet loss and energy consumption in sensor networks depend critically on the quality of the network's wireless links. In turn, link quality depends on the environment in which the RF signals propagate and the locations of the link's endpoints. Experimental results have shown that a low-power wireless link can be in one of three states or 'regions', as the inter-node distance increases: connected, transitional (gray), and disconnected. The gray region is characterized by extreme variability, whereby small differences in distance or endpoint locations can lead to pronounced differences in loss rates. However, not all is lost. This work investigates the spatial characteristics of the gray region and experimentally shows that one can efficiently identify links with low loss rates within the radio's gray region. One of the possible applications of this finding is in the design of sparse, yet low-loss network deployments.
The graph on the right shows the measured variations in Packet Reception Ratio (PRR) over a two-dimensional grid in a parking lot. Red cells correspond to coverage holes, while Green cells designate good receptions (i.e., PRR > 85%). The X- and Y-axis values denote relative distances to the transmitter. Significant local variations are evident from this graph, demonstrating the spatial characteristics of the radio's Gray Region.
A preliminary version of this work appeared as a poster at IPSN 2009: Download PDF
The whole dataset is available on this page.