Multi-prediction particle filter for efficient parallelized implementation

Abstract

Particle filter (PF) is an emerging signal processing methodology, which can effectively deal with nonlinear and non-Gaussian signals by a sample-based approximation of the state probability density function. The particle generation of the PF is a data-independent procedure and can be implemented in parallel. However, the resampling procedure in the PF is a sequential task in natural and difficult to be parallelized. Based on the Amdahl's law, the sequential portion of a task limits the maximum speed-up of the parallelized implementation. Moreover, large particle number is usually required to obtain an accurate estimation, and the complexity of the resampling procedure is highly related to the number of particles. In this article, we propose a multi-prediction (MP) framework with two selection approaches. The proposed MP framework can reduce the required particle number for target estimation accuracy, and the sequential operation of the resampling can be reduced. Besides, the overhead of the MP framework can be easily compensated by parallel implementation. The proposed MP-PF alleviates the global sequential operation by increasing the local parallel computation. In addition, the MP-PF is very suitable for multi-core graphics processing unit (GPU) platform, which is a popular parallel processing architecture. We give prototypical implementations of the MP-PFs on multi-core GPU platform. For the classic bearing-only tracking experiments, the proposed MP-PF can be 25.1 and 15.3 times faster than the sequential importance resampling-PF with 10,000 and 20,000 particles, respectively. Hence, the proposed MP-PF can enhance the efficiency of the parallelization.

Fast, parallel implementation of particle filtering on the GPU architecture

“…For the measurements, they used a bearing-only tracking (BOT) model with 25 time steps; therefore, we can make direct comparison for the estimation error. For [12], the position error is in the range of 0.06245 to 0.06226, which is slightly lower than our error, but still the same range. Execution time, which is the sum of sampling, weight normalization and resampling times in [12], and total runtime in our work (including memory transfers, file I/O, etc.…”

“…For [12], the position error is in the range of 0.06245 to 0.06226, which is slightly lower than our error, but still the same range. Execution time, which is the sum of sampling, weight normalization and resampling times in [12], and total runtime in our work (including memory transfers, file I/O, etc. ), shall be compared to our proposed algorithm with regard to the different devices, which still indicates that our technique is faster (see Table 1).…”

“…There have been some former implementations to parallel architectures [10][11][12][13][14][15][16][17][18][19]. In [11], an implementation strategy is proposed which is parallel; however, it cannot maintain the local connections of the particles.…”

“…However, there is a trade-off between particle number and estimation accuracy. The spreadingnarrowing technique in [12] proposes a solution. A tolerable N number of basis particles generates an N × P large set by propagating each particle based on the system transition model for a sequence of P states.…”