Parallel Bayesian inference of range and reflectance from LaDAR profiles

• Published in: Journal of parallel and distributed computing 2013-04, Vol.73 (4), p.383-399
• Main Authors: Ye, Jing, Wallace, Andrew M., Al Zain, Abdallah, Thompson, John
• Format: Article
• Language: English
• Subjects: Applied sciences; Beam characteristics: profile, intensity, and power; spatial pattern formation; Computer science; control theory; systems; Computer systems and distributed systems. User interface; Data parallelism; Distributed parallel computation; Dynamic task allocation; Exact sciences and technology; Full waveform; Fundamental areas of phenomenology (including applications); Laser optical systems: design and operation; Laser radar; Optics; Physics; Reversible jump Markov chain Monte Carlo; Software; Varying-dimensional signal analysis
• ISSN: 0743-7315; 1096-0848
• DOI: 10.1016/j.jpdc.2012.12.003

Bayesian analysis using reversible jump Markov chain Monte Carlo (RJMCMC) algorithms improves the measurement accuracy, resolution and sensitivity of full waveform laser detection and ranging (LaDAR), but at a significant computational cost. Parallel processing has the potential to significantly reduce the processing time, but although there have been several strategies for Markov chain Monte Carlo (MCMC) parallelization, adaptation of these strategies to RJMCMC may degrade parallel performance. In this paper, we describe an approach to parallel RJMCMC processing that combines data and sampling parallelism in a single framework. This approach, Data Parallel State Space Decomposed RJMCMC (DP SSD-RJMCMC), can be adapted to different parallel cluster size, improve sampling efficiency and maintain parameter estimation accuracy. Formally, it forms a group of parallel chains by decomposing the state space into subsets of parameter space. Each subset has different but restricted dimensionality, and is assigned with an independent chain of variable length. To further improve load balancing, we also employ data decomposition, forming a task queue and conducting dynamic task allocation. The MPI-based implementation on a 32-node Beowulf cluster leads to significant speedup, typically of the order of 15–25 times, while maintaining the estimation accuracy. ► First implementation of a parallel MIMD algorithm for full waveform Lidar analysis. ► A method to parallelize RJMCMC chains whose length is controlled by convergence diagnostics. ► Combined functional and data decomposition for improved load balancing. ► Demonstrable results on both simulated and real data using our own LiDAR technology. ► Comparable accuracy combined with significant speedup on a Beowulf MIMD architecture.