In this paper, a particular class of block-circulant low-density parity-check (BC-LDPC) codes referred to as cylinder-type BC-LDPC (CTBC-LDPC) codes is studied. We represent a cylinder-type parity-check matrix H by a graph called the block-structure graph of H and denoted by BSG(H). Using the properties of BSG(H) we show that CTBC matrices with column-weight two and girth an arbitrary multiple of 8 can be constructed, while for a CTBC matrix H with column-weight ℓ ≥ 3 this girth cannot exceed 12. An algorithm generating CTBC-LDPC codes of arbitrary possible girth is given. The algorithm produces a large group of codes with flexible rate, length and girth. From performance perspective over AWGN channel, the maximum girth-12 CTBC-LDPC codes with column-weight at least three generated by the given algorithm are at least as good as their random-like counterpart codes and outperform the codes of the same size constructed by successive-level-growth, progressive-edge-growth, and partition-and-shift methods.

Four classes of maximum-girth geometrically structured column-weight-two regular quasi-cyclic (QC) low-density parity-check (LDPC) codes are introduced. Two classes of these codes, referred to as Type-I and Type-II codes, are with row-weights 4 and 3, and maximum girths 16 and 24, respectively. The idea behind the construction of these two classes of codes, with rates at least 1/2 and 1/3, is slightly generalized to obtain two classes of variable-high-rate codes, referred to as Type-III1 and Type-III2 codes, with maximum girth 20 and 16, respectively. A low-complexity deterministic algorithm for constructing these four classes of codes is given. The algorithm generates maximum-girth Type-I and Type-II codes with almost arbitrary length eta not less than 216 and 243, respectively. The output of the algorithm substantially improves on some of the previously best known codes constructed using a randomized progressive edge-growth (RPEG) algorithm. For instance, we have rate-0.71 Type-III1 codes of lengths 308 and 728 with girths 10 and 12, respectively, versus the code lengths 385 and 840 obtained by the RPEG algorithm. Simulation results on AWGN channel confirm that, from BER performance perspective, the constructed LDPC codes are superior to the column-weight-two LDPC codes constructed by the previously reported methods. The generator matrix G(D) of the convolutional codes associated with Type-I and Type-II codes is given. The free distance dfreeof such a convolutional code is equal to the minimum distance of the corresponding QC block code.

A class of maximum-girth geometrically structured regular (n, 2, k≥5) (column-weight 2 and row-weight k) quasi-cyclic low-density parity-check (LDPC) codes is presented. The method is based on cylinder graphs and the slope concept. It is shown that the maximum girth achieved by these codes is 12. A low-complexity algorithm producing all such maximum-girth LDPC codes is given. The shortest constructed code has a length of 105. The minimum length n of a regular (2, k) LDPC code with girth g=12 determined by the Gallager bound has been achieved by the constructed codes. From the perspective of performance these codes outperform the column-weight 2 LDPC codes constructed by the previously reported methods. These codes can be encoded using an erasure decoding process.

This paper introduces an adaptive growing and pruning radial basis function (GAP-RBF) neural network for on-line identification of hybrid systems. The main idea is to identify a global nonlinear model that can predict the continuous outputs of hybrid systems. In the proposed approach, GAP-RBF neural network uses a modified unscented kalman filter (UKF) with forgetting factor scheme as the required on-line learning algorithm. The effectiveness of the resulting identification approach is tested and evaluated on a simulated benchmark hybrid system.

In this paper we propose a procedure for recursive identification of discrete time piecewise affine (PWA) hybrid systems. The PWA identification problem involves the estimation of both the parameters of affine submodels and the partition of the PWA map from data. In this paper the submodel parameters estimation problem is solved via recursive k-plane clustering algorithm and the problem of region estimation is performed by incremental proximal support vector machine. Also, the effect of mode switching on the estimated parameters convergence is investigated

A new bounded-error approach for the identification of discrete time hybrid systems in the piece-wise affine (PWA) form is introduced. The PWA identification problem involves the estimation of the number of affine submodels, the parameters of affine submodels and the partition of the PWA map from data. By imposing a bound on the identification error, we formulate the PWA identification problem as a MIN PFS problem (partition into a minimum number of feasible subsystems) and propose a greedy clustering-based method for tackling it. The proposed approach yields to better results than the greedy randomized relaxation algorithm used in previous methods. Also, it is not sensitive to the overestimation of model orders and changes in the tuning parameters and therefore finding a right combination of the tuning parameters of the algorithm to get a model with prescribed bounded prediction error is simple