An Original Correction Method for Indoor Ultra Wide Band Ranging-Based Localisation System

Abstract

Abstract. During this decade, Wireless Sensor Networks (WSNs) brought an increasing interest in the industrial and research world. One of their applications is the indoor localization. The ranging, i.e. the distance evaluation mechanism between nodes, is required to determine the position of the nodes. The research work presented in this article aims to use Ultra Wide Band (UWB) radio links to achieve an efficient ranging, based on Time of Flight (ToF) measurement. A good solution consists in integrating ranging traffic into the usual network messages. However, the ToF ranging process is based on information exchanges which are temporally constrained. Once this information is encapsulated into the usual messages, the temporal constraint cannot be honoured, resulting in important ranging errors due to clock drifts. To mitigate these errors, we have introduced an original dynamic correction technique which enables a precision of twenty centimetres allowing the inclusion of ranging traffic in usual traffic.

N-TWR: An accurate time-of-flight-based N-ary ranging protocol for Ultra-Wide band

“…The first conclusion we can draw from these results is that the estimation of the ToF is significantly improved when compensating it with the skew estimation. This result was also confirmed in [26] for a skew estimated by means of the DecaWave's functionality. Secondly, we can see that there is no significant difference between the estimation done by both skew compensation approaches.…”

“…Advanced hardware functionalities are for instance (i ) the ability to learn the clock skew between two nodes engaged in a ranging operation as leveraged in [26] , or ii ) the ability to schedule the emission of a frame at a specific future date as leveraged in [8] . If such features are available, it is possible to improve the precision of the protocols proposed herein or reduce the number of messages required.…”

“…Hence, the time difference t 3 − t 2 measured with the clock rate of A and t 4 − t 1 with the clock rate of B are really different, even though they represent almost the same time interval. Authors in [26] propose a skew compensation based on a DecaWave DW10 0 0 functionality that allows obtaining the frequency relationship between nodes:…”

“…Successive message exchanges will allow node A to estimate a more accurate skew by means of a linear regression approach which will consider, not only, the current points ( t 2 , t 1 ) and ( t 3 , t 4 ), but also those previously computed. By successively computing the slope of the regression line, the estimation of the ToF can be improved in the same way as done in [26] but using the slope learned with the linear regression method:…”

Error Corrections for Ultrawideband Ranging

“…The residual error is now independent of the response time, and only the time-of-flight T CCSS is the limiting factor. This technique is widely used in the community [18], [19], [20]. The reason why we will not use this method later on is that in [13], we showed that the integrator of the PLL depends on the signal power, and therefore introduces additional inaccuracies.…”