Calibration and Measurement of Signal Strength for Sensor Localization

Calibration and Measurement of Signal Strength for Sensor Localization

Neal Patwari (University of Utah, USA) and Piyush Agrawal (University of Utah, USA)
DOI: 10.4018/978-1-60566-396-8.ch005

Abstract

A number of practical issues are involved in the use of measured received signal strength (RSS) for purposes of localization. This chapter focuses on device effects and modeling problems which are not well covered in the literature, such as transceiver device manufacturing variations, battery effects on transmit power, nonlinearities in RSSI circuits, and path loss model parameter estimation. The authors discuss both the negative impacts of these effects and inaccuracies, and adaptations used by particular localization algorithms to be robust to them, without discussing any algorithm in detail. The authors present measurement methodologies to characterize these effects for wireless sensor nodes, and report the results from several calibration experiments to quantify each discussed effect and modeling issue.
Chapter Preview
Top

Introduction

Signal-strength based localization can be deceptively simple. Receivers are generally capable of measuring and reporting to higher layers information about received signal strength, so it can seem like it should be easy to take these measurements and use them directly in a localization algorithm. Significant research has developed algorithms for localization, assuming that measured signal strength has already been converted into distance estimates, and little research discusses the details of how to perform those conversions.

Multipath fading in radio channels is universally regarded to be the main degradation to RSS-based location estimates, and rightfully so – significant shadowing and small-scale fading caused by the channel is largely unavoidable and unpredictable. Beyond that, however, there can be severe degradations caused by a lack of understanding of the non-idealities of the measurement process, and inaccurate knowledge of channel parameters. If RSS-based localization is to be attempted, a designer must be able to characterize and cope with these non-idealities and imperfect knowledge.

This chapter is written to present real-world calibration and non-linearity problems in RSS measurements and how to deal with them. We follow RSS-based localization from the transmitter to the receiver, and in multiple stages in the receiver, as shown in Figure 1. The intended audience is anyone who intends to implement or has already implemented RSS-based localization algorithms which are to operate well in real-world deployments. We present our work in RSS-based localization algorithms only briefly. We have found that the experience of accurately using measured signal strength, in general, is as challenging as the localization algorithm itself.

Figure 1.

RSS-based localization requires characterization of both the transmitter and receiver, and the ability to convert measured RSSI into path loss prior to input into a localization algorithm. Path loss estimation requires knowing transmitter parameters and may require feedback to the transmitter to control its transmit power

978-1-60566-396-8.ch005.f01

Our chapter is organized as follows. First, in section Propagation Effects, we relate some of the literature on path loss models as a function of distance and the effects of shadowing and multipath fading. Then, in section Device Effects, we discuss a method for accurately characterizing transmit power as a function of device settings and battery voltage, and receiver RSSI values as a function of the particular device performing the measurement. The transmit power characteristics are necessary to translate measured RSS into accurate path loss values. The receiver characterization reveals the details of the nonlinearities in measured RSSI. Then, section Channel Experiments with Power Control describes a protocol and algorithm for transmit power control, to avoid RSSI saturation without sacrificing node connectivity. Thirdly, the section titled Ranging Using Measured Path Loss discusses the conversion of path loss, calculated using the results of the Device Effects section, into an estimate of range. Finally, a section called RSS-Based Localization Algorithms discusses how the lessons discussed in this chapter apply in our RSS-based location algorithm implementation.

Top

Propagation Effects

Multipath fading, shadowing, and antenna effects cause great variations in the measured RSS in real-world environments, degrading its ability to produce accurate distance or position estimates. This section serves to emphasize these well-reported effects in order to position the importance of studying device effects on measured RSS. As we delve deeper into device effects which cause RSS errors, we will be able to position them in context to the larger problem of RSS-based position estimation.

Path losses, on average, increase with distance – the increase is due to “large-scale” path loss (Hashemi, 1993), which are proportional to where is a path loss exponent, and d is the path length. But the path loss between two radios at particular positions is very much a function of the objects in the environment between them and the position and orientation of the antennas. Movement on the order of centimeters or changing channel from one frequency to another can cause dramatic path loss differences because of “multipath fading” or “small-scale fading”.

Complete Chapter List

Search this Book:
Reset