Event-driven distributed filtering method and system for microphone network

Abstract

The invention discloses an event-driven distributed filtering method and system for a microphone network. The method comprises the following steps: constructing a distributed acoustic array sensor network; establishing a topology communication link structure between microphone nodes, and setting an event trigger condition threshold ; Determine each initial value of the initial state; for time k, each microphone obtains its own latest state estimate, current state estimate and actual error covariance; set an event-driven trigger mechanism; build a sound source at time k+1 The state estimation equation of the target, update the state estimation value in combination with the event trigger mechanism; update the upper bound of the minimum error covariance; use the upper bound of the minimum error covariance to solve the measurement gain and the state error gain in the state estimation equation, and carry out the sound source target. State estimation. The invention can improve the information utilization efficiency of the microphone node and realize the accurate tracking of the sound source target.

Description

technical field [0001] The invention relates to the technical field of sound source target tracking, in particular to an event-driven microphone network distributed filtering method and system. Background technique [0002] Acoustic array sensor network target tracking technology is a technology to track targets based on sound source information. Acoustic array sensor network involves the mutual communication between multiple microphone nodes, and generally adopts the processing method of distributed Kalman filtering. However, most of the existing distributed acoustic array sensor networks implicitly adopt a periodic time-driven strategy, where a microphone propagates its estimated value to its neighboring microphones at each moment. , the entire sensor network requires a lot of communication costs. Moreover, this periodic time-driven strategy may lead to unnecessary microphone information transmission, thus wasting the bandwidth resources of the acoustic array sensor netw...

AI Technical Summary

Problems solved by technology

However, most of the existing distributed acoustic array sensor networks implicitly employ a periodic time-driven strategy, where a microphone propagates its estimate to its neighbors at each instant, thus, due to the frequent communication of , the entire sensor network requires a large communication cost

Moreover, this periodic time-driven strategy may lead to unnecessary microphone information transmission, which wastes the bandwidth resources of the acoustic array sensor network.

In addition, frequent information communication among sensor nodes may cause severe network congestion, which further leads to packet loss or communication delay in the network

Images