arXiv subjects

Paul H. Siegel

Publications and source records attributed to Paul H. Siegel.

65 records · Page 4Linked to original sources

arXiv2008

Gaussian Belief Propagation Solver for Systems of Linear Equations

The canonical problem of solving a system of linear equations arises in numerous contexts in information theory, communication theory, and related fields. In this contribution, we develop a solution based upon Gaussian belief propagation (GaBP) that does not involve direct matrix inversion. The iterative nature of our approach allows for a distributed message-passing implementation of the solution algorithm. We also address some properties of the GaBP solver, including convergence, exactness, its max-product version and relation to classical solution methods. The application example of decorrelation in CDMA is used to demonstrate the faster convergence rate of the proposed solver in comparison to conventional linear-algebraic iterative solution methods.

cs.IT↗

arXiv2008

Gaussian Belief Propagation Based Multiuser Detection

In this work, we present a novel construction for solving the linear multiuser detection problem using the Gaussian Belief Propagation algorithm. Our algorithm yields an efficient, iterative and distributed implementation of the MMSE detector. We compare our algorithm's performance to a recent result and show an improved memory consumption, reduced computation steps and a reduction in the number of sent messages. We prove that recent work by Montanari et al. is an instance of our general algorithm, providing new convergence results for both algorithms.

cs.IT↗

arXiv2008

Gaussian Belief Propagation for Solving Systems of Linear Equations: Theory and Application

The canonical problem of solving a system of linear equations arises in numerous contexts in information theory, communication theory, and related fields. In this contribution, we develop a solution based upon Gaussian belief propagation (GaBP) that does not involve direct matrix inversion. The iterative nature of our approach allows for a distributed message-passing implementation of the solution algorithm. We address the properties of the GaBP solver, including convergence, exactness, computational complexity, message-passing efficiency and its relation to classical solution methods. We use numerical examples and applications, like linear detection, to illustrate these properties through the use of computer simulations. This empirical study demonstrates the attractiveness (e.g., faster convergence rate) of the proposed GaBP solver in comparison to conventional linear-algebraic iterative solution methods.

cs.IT↗

arXiv2007

Single-Exclusion Number and the Stopping Redundancy of MDS Codes

For a linear block code C, its stopping redundancy is defined as the smallest number of check nodes in a Tanner graph for C, such that there exist no stopping sets of size smaller than the minimum distance of C. Schwartz and Vardy conjectured that the stopping redundancy of an MDS code should only depend on its length and minimum distance. We define the (n,t)-single-exclusion number, S(n,t) as the smallest number of t-subsets of an n-set, such that for each i-subset of the n-set, i=1,...,t+1, there exists a t-subset that contains all but one element of the i-subset. New upper bounds on the single-exclusion number are obtained via probabilistic methods, recurrent inequalities, as well as explicit constructions. The new bounds are used to better understand the stopping redundancy of MDS codes. In particular, it is shown that for [n,k=n-d+1,d] MDS codes, as n goes to infinity, the stopping redundancy is asymptotic to S(n,d-2), if d=o(\sqrt{n}), or if k=o(\sqrt{n}) and k goes to infinity, thus giving partial confirmation of the Schwartz-Vardy conjecture in the asymptotic sense.

cs.IT↗

arXiv2007

Graph-Based Decoding in the Presence of ISI

We propose an approximation of maximum-likelihood detection in ISI channels based on linear programming or message passing. We convert the detection problem into a binary decoding problem, which can be easily combined with LDPC decoding. We show that, for a certain class of channels and in the absence of coding, the proposed technique provides the exact ML solution without an exponential complexity in the size of channel memory, while for some other channels, this method has a non-diminishing probability of failure as SNR increases. Some analysis is provided for the error events of the proposed technique under linear programming.

cs.IT↗

arXiv2007

Adaptive Methods for Linear Programming Decoding

Detectability of failures of linear programming (LP) decoding and the potential for improvement by adding new constraints motivate the use of an adaptive approach in selecting the constraints for the underlying LP problem. In this paper, we make a first step in studying this method, and show that it can significantly reduce the complexity of the problem, which was originally exponential in the maximum check-node degree. We further show that adaptively adding new constraints, e.g. by combining parity checks, can provide large gains in the performance.

cs.IT↗

arXiv2006

Adaptive Linear Programming Decoding

Detectability of failures of linear programming (LP) decoding and its potential for improvement by adding new constraints motivate the use of an adaptive approach in selecting the constraints for the LP problem. In this paper, we make a first step in studying this method, and show that it can significantly reduce the complexity of the problem, which was originally exponential in the maximum check-node degree. We further show that adaptively adding new constraints, e.g. by combining parity checks, can provide large gains in the performance.

cs.IT↗

arXiv2005

Improved Upper Bounds on Stopping Redundancy

Let C be a linear code with length n and minimum distance d. The stopping redundancy of C is defined as the minimum number of rows in a parity-check matrix for C such that the smallest stopping sets in the corresponding Tanner graph have size d. We derive new upper bounds on the stopping redundancy of linear codes in general, and of maximum distance separable (MDS) codes specifically, and show how they improve upon previously known results. For MDS codes, the new bounds are found by upper bounding the stopping redundancy by a combinatorial quantity closely related to Turan numbers. (The Turan number, T(v,k,t), is the smallest number of t-subsets of a v-set, such that every k-subset of the v-set contains at least one of the t-subsets.) We further show that the stopping redundancy of MDS codes is T(n,d-1,d-2)(1+O(n^{-1})) for fixed d, and is at most T(n,d-1,d-2)(3+O(n^{-1})) for fixed code dimension k=n-d+1. For d=3,4, we prove that the stopping redundancy of MDS codes is equal to T(n,d-1,d-2), for which exact formulas are known. For d=5, we show that the stopping redundancy of MDS codes is either T(n,4,3) or T(n,4,3)+1.

cs.IT↗

arXiv2005

Coding for the Optical Channel: the Ghost-Pulse Constraint

We consider a number of constrained coding techniques that can be used to mitigate a nonlinear effect in the optical fiber channel that causes the formation of spurious pulses, called ``ghost pulses.'' Specifically, if $b_1 b_2 ... b_{n}$ is a sequence of bits sent across an optical channel, such that $b_k=b_l=b_m=1$ for some $k,l,m$ (not necessarily all distinct) but $b_{k+l-m} = 0$, then the ghost-pulse effect causes $b_{k+l-m}$ to change to 1, thereby creating an error. We design and analyze several coding schemes using binary and ternary sequences constrained so as to avoid patterns that give rise to ghost pulses. We also discuss the design of encoders and decoders for these coding schemes.

cs.IT↗

arXiv2005

On the Asymptotic Performance of Iterative Decoders for Product Codes

We consider hard-decision iterative decoders for product codes over the erasure channel, which employ repeated rounds of decoding rows and columns alternatingly. We derive the exact asymptotic probability of decoding failure as a function of the error-correction capabilities of the row and column codes, the number of decoding rounds, and the channel erasure probability. We examine both the case of codes capable of correcting a constant amount of errors, and the case of codes capable of correcting a constant fraction of their length.

cs.IT↗

arXiv2005

Relaxation Bounds on the Minimum Pseudo-Weight of Linear Block Codes

Just as the Hamming weight spectrum of a linear block code sheds light on the performance of a maximum likelihood decoder, the pseudo-weight spectrum provides insight into the performance of a linear programming decoder. Using properties of polyhedral cones, we find the pseudo-weight spectrum of some short codes. We also present two general lower bounds on the minimum pseudo-weight. The first bound is based on the column weight of the parity-check matrix. The second bound is computed by solving an optimization problem. In some cases, this bound is more tractable to compute than previously known bounds and thus can be applied to longer codes.

cs.IT↗