Abstract: An apparatus and method for classifying a driver's activity state use sensor data relating to at least one vehicle operating condition to determine driver activity and classify into one of at least two states. A two state classifier (200) uses inputs related to vehicle monitors (210), driving conditions (212), driver activity (214), vehicle controls (216), and driver identification (218) to set an output (204) to a status corresponding to “maneuver” and “non-maneuver.

Abstract: An apparatus and method for classifying a driver's activity state use sensor data relating to at least one vehicle operating condition to determine driver activity and classify into one of at least two states. A two state classifier (200) uses inputs related to vehicle monitors (210), driving conditions (212), driver activity (214), vehicle controls (216), and driver identification (218) to set an output (204) to a status corresponding to “maneuver” and “non-maneuver.

Abstract: Methods and apparatus for recovering a desired signal from a composite signal. In one method, sample-based information maximization techniques for blind deconvolution are utilized to learn a filter (22) which removes undesired dependencies from the composite signal. Entropy as an expectation over individual samples of the composite signal is maximized. In another method, dependencies of the composite signal are selectively removed by learning a first filter (42) that removes the desired dependencies and outputs a filtered signal. The filtered signal is used as the learning material for a second filter (44) such that the second filter learns to remove only the remaining dependencies (undesired dependencies) from the filtered signal. The composite signal is applied to the second filter (44) which thereby removes the undesired dependencies.

Abstract: Multiple digital radio signals transmitted simultaneously at the same frequency are separated and the source signals recovered using blind source separation techniques. RF signals are received at a communications station (20) and downconverted to baseband signals. Blind Source Separation (BSS) based processes are applied to the baseband signals prior to demodulation to recover the source signals. To recover the source signals, a probability density function (PDF) model of the baseband source signal is generated (step 12) and an adaptation equation produced in accordance with the particular PDF model and BSS process (step 14). Using the known PDFs provides quicker adaptation to dynamic conditions. A separation matrix is learned by applying the adaptation equation to the baseband signals (step 16) and applied to the baseband signals to recover the source signals (step 18).

Abstract: A method of recovering source signals from an equal number of mixed signals including components of the source signals. Channel one and two mixed signals are multiplied by channel one and two adaptive weights to produce channel one and two product signals, respectively. A channel one filtered feedback signal is generated from a channel two approximation signal and a channel two filtered feedback signal is generated from a channel one approximation signal. The channel one filtered feedback signal is added to the channel one product signal to produce the channel one approximation signal and the channel two filtered feedback signal is added to the channel two product signal to produce the channel two approximation signal. A nonlinear function is applied to the approximation signals to produce output signals. The adaptive weights and the filtered feedback signals are adjusted to maximize entropy of the output signals.