Neural Network Assisted Identification of the Absence of Direct Path in Indoor Localization

Time of Arrival (TOA) based indoor positioning systems are considered to be the high precision alternatives to other positioning systems employing received signal strength (RSS) or Angle of Arrival (AOA). However, such systems suffer from the blockage of the direct path (DP) and occurrence of undetected direct path (UDP) condition and their performance degrades drastically in such conditions. Erroneous detection of the other multipath components (MPCs) as DP, which is the indicator of the true distance between the transmitter and the receiver, will introduce substantial ranging and localization errors into the system. Therefore, identification of the occurrence of large ranging errors and absence of DP from the received radio signal is our subsequent concern. After identification, the second step is to remedy the ranging errors in such UDP conditions. In this paper we present a methodology, based on an application of artificial neural network (ANN) design, to identify the UDP conditions and mitigate the ranging error using statistics extracted from wideband frequency domain indoor measurements conducted in a typical office building. The system bandwidth used for the frequency domain measurement was 500 MHz centered around 1 GHz.