The two-path tree search is a detection algorithm that improves the performance of standard decision-feedback equalizer detectors by examining two likely paths through the decision tree. The method achieves about 1-dB improvement over standard decision-feedback equalization for typical magnetic recording channels used in hard disk drives. The implementation complexity is modest and is compared with other improved decision-feedback structures.

The reasons for pursuing analog implementations of the Viterbi algorithm for detection are presented. Some of the tradeoffs to be considered in choosing an analog architecture are discussed, and a suitable architecture for investigation is given. Simulation results based on the chosen architecture are presented and clearly show that analog implementations of the maximum likelihood sequence detector are worth pursuing. In addition, the simulation results provide quantitative guidelines for establishing the required specifications of the individual circuit blocks required to construct a Viterbi detector.>

A neural network has been developed for use in local closed-loop gripper control using a tactile sensor array. Backpropagation is used to train a network with a single hidden layer to recognize the angle of contact between a cylindrical finger with an embedded sensor array and a cylindrical rod. Simulation results are presented and discussed.>