Your Location Within Site:
Subspace methods
Industrial Research is developing methods for classification and separation of signals based on higher dimensional signal representation. Applications include enhanced resolution, separation, communication or location estimation in acoustic or wireless reverberant situations.
Representation of signals, mixtures of signals and reverberant combinations of signals as higher dimensional signals has provided some powerful algorithmic solutions to problems such as estimation, identification or location. Existing algorithms applied as spectrum estimation include MUSIC, ESPRIT and many variations. The theory and algorithms involved is an active research area with the number of new applications and methods of application growing.
Subspace methods offer promise to difficult tasks such a wireless location in reverberant locations. Subspace methods provide some ability to separate direct paths from the scattered paths for such as direction finding.
Alternatively the nature of the higher-dimensional subspace of a signal provides a “subspace signature” for the signal itself and therefore suggests algorithms for location that combine separation and classification as two different treatments of the same signals. The convergence of these methods and pattern recognition algorithms into a single framework leads to improved subspace classification techniques.
Our current application areas, in implementation, include reverberant propagation for indoor identification and location and enhanced resolution for sonar systems, particularly in reverberant condition.
Industrial Research Limited has capability to apply these techniques to development of algorithms for many applications. Our expertise for implementation in software, DSP[?] and FPGA[?] provides a capability to transfer these technologies into industrial applications.