.png){kind=icon}
Return
Accelerated Consensus-Based SPSA Algorithm for Multisensor Multitarget Tracking Problem
.png){kind=icon}
PRE
AI
DOI：10.1109/tac.2026.3678473.png)
.png){kind=icon}
Original request for help.png){kind=icon}
Share.png){kind=icon}
Save
Abstract
En 中文
Real-time control of sensor networks under communication constraints has broad applications, including target tracking and mobile robotics. Traditional centralized methods face congestion and delay issues as system size grows, motivating a shift toward decentralized multiagent approaches. This article studies a nonstationary mean-risk optimization model for distributed sensor networks with distance-only noisy measurements and unknown-but-bounded disturbances. An accelerated consensus-based simultaneous perturbation stochastic approximation (A-SPSA) algorithm is proposed and analyzed under time-varying communication graphs and noisy information exchange. The method relies exclusively on zeroth-order oracle feedback and admits explicit error residual and convergence rate guarantees. Numerical experiments in a target-tracking scenario illustrate the improved convergence behavior of the accelerated scheme compared to a baseline distributed SPSA.
Keywords:
Optimization
Noise measurement
Noise
Robot sensing systems
Heuristic algorithms
Convergence
Target tracking
Stochastic processes
Real-time systems
Estimation
Distributed sensor networks
multiagent systems
Nesterov acceleration
stochastic optimization
target tracking
Journal
IF:
7
Papers:
1.3W
Citations:
6.7W
Organization
Cited Papers
No cited papers available