Improved Particle Filter Resampling Architectures

Abstract

The most challenging aspect of particle filtering hardware implementation is the resampling step. This is because of high latency as it can be only partially executed in parallel with the other steps of particle filtering and has no inherent parallelism inside it. To reduce the latency, an improved resampling architecture is proposed which involves pre-fetching from the weight memory in parallel to the fetching of a value from a random function generator along with architectures for realizing the prefetch technique. This enables a particle filter using M particles with otherwise streaming operation to get new inputs more often than 2M cycles as the previously best approach gives. Results show that a pre-fetch buffer of five values achieves the best area-latency reduction trade-off while on average achieving an 85% reduction in latency for the resampling step leading to a sample time reduction of more than 40%. We also propose a generic division-free architecture for the resampling steps. It also removes the need of explicitly ordering the random values for efficient multinomial resampling implementation. In addition, on-the-fly computation of the cumulative sum of weights is proposed which helps reduce the word length of the particle weight memory. FPGA implementation results show that the memory size is reduced by up to 50%.

Zaman Serisi Veri Kümeleri İçin Olasılığa Dayalı Tahmin Yöntemi

“…A resampling phase [16] will be applied as shown in the following pseudocode, Figure (2). In this case, and according to its normalized weight, each particle gives its corresponding successors at moment (t + 1).…”

Aging Time Prediction Model Analysis and Numerical Simulation of Random Degradation Equipment Based on Big Data Linkage Technology

“…e remaining life of the equipment is predicted based on the condition monitoring data. Typical parameter identification and updating methods include Kalman filtering [11,12], particle filtering [13,14], and Bayesian methods [15][16][17].…”

“…The remaining life of the equipment is predicted based on the condition monitoring data. Typical parameter identification and updating methods include Kalman filtering [ 11 , 12 ], particle filtering [ 13 , 14 ], and Bayesian methods [ 15 – 17 ]. The common mechanistic models used for the remaining life prediction include the Paris model, Forman model, and various improvements and extensions based on them, mainly to describe the crack expansion and laminar crack growth [ 18 ].…”

FPGA Implementation of Particle Filters for Robotic Source Localization

“…Although the state vector estimation and likelihood computations are parallelized, the re-sampling step is done sequentially, which is the major drawback of the architecture. Recently, Alam et al [18] proposed an improved re-sampling architecture by introducing a weight prefetch mechanism to reduce the latency of the re-sampling step. In this technique, new particle weights are pre-fetched along with the random values concurrently, which help in reducing the total number of cycles for re-sampling.…”

“…Mountney et al[17] implemented a Bayesian auxiliary particle filter algorithm (BAPF) on an FPGA for brain machine interfaces with a 90 kHz sampling rate. Alam et al[18] proposed an improved multinomial re-sampling scheme to reduce the re-sampling latency and implemented the same on an FPGA with 299 LUTs and 2 BRAMs for 1k particles. However, they don't report the sampling rate of the whole architecture.…”