Assisted-GPS-based Snap-shot GPS Receiver with FFT-accelerated Collective Detection: Time Synchronisation and Search Space Analysis

摘要

In this paper, the architecture of the Assisted-GPSbased snap-shot GPS receiver is introduced. Such asystem would be suitable for emergency services,freeing up the processing burden of the clienthandset receiver while enhanced signal processingalgorithms are used to process a “snap-shot” oftransferred raw radio-frequency data at an A-GPSserver. This system can be treated as a hybrid ofadvanced technologies that comprises the conceptsof Assisted-GPS, snap-shot GPS receiver andCollective Detection. The implementation and timesynchronisation issues of such a system are the mainfocus of this paper.Specifically, two implementations are described andcontrasted by means of computational efficiency andimplementation losses. Subsequent investigationsthen focus on the resulting position accuracy andprecision due to non-ideal time synchronisationbetween the Base Station and the Mobile Station.The findings highlight the importance of consideringtime synchronisation errors (i.e. coarse time error)when evaluating the performance of CollectiveDetection. Also, the effects of the use of variousspatial step sizes with non-zero coarse time errors are shown. The associated precisions are empiricallycharacterised as a function of time synchronisationerror and position-clock domain step size.Finally, two computationally efficient algorithms areproposed and analysed in this paper to compensatefor the time synchronisation issues. One of themethods explored is shown to have improvedconvergence and is capable of resolving coarse timeerror to a high resolution for a relatively lowincrease in computational load.There is a trade-off between precision andcomputational load by way of step size selection,which is a parameter in the Collective Detectionalgorithm. To improve precision without increasingthe instantaneous computational load, a simpleaveraging approach is applied to multiple processedsnap-shots. This approach is especially applicablewhen Collective Detection is operating under limitedcomputational resource or limited receiver-to-servernetwork bandwidth.Unlike many other previous research contributions,in all investigations, the entire Common Clock Biassearch range (i.e. 0-300km) is considered in theCollective Detection search space. In addition, livesignals are used in real-life non-ideal scenarios forperformance evaluations.