Enrico Paolini

Generalized IRA Erasure Correcting Codes for Hybrid Iterative/Maximum Likelihood Decoding

The design of low-density parity-check (LDPC) codes under hybrid iterative / maximum likelihood decoding is addressed for the binary erasure channel (BEC). Specifically, we focus on generalized irregular repeat-accumulate (GeIRA) codes, which offer both efficient encoding and design flexibility. We show that properly designed GeIRA codes tightly approach the performance of an ideal maximum distance separable (MDS) code, even for short block sizes. For example, our (2048,1024) code reaches a codeword error rate of 10-5 at channel erasure probability isin= 0.450, where an ideal (2048,1024) MDS code would reach the same error rate at isin = 0.453.

Target detection metrics and tracking for UWB radar sensor networks

A radar sensor network (RSN) is a wireless network system, typically composed of one transmitting (TX) node and several receiving (RX) nodes, aimed at detecting and tracking an intruder (target) moving within a given surveillance area. In this paper, RSNs based on impulse radio ultra-wideband (UWB) are investigated. In the considered system, at each scanning each RX node calculates a soft image of the surveillance area based on the target-scattered UWB pulses. All images are then transferred to a fusion node where the decision about target presence or absence is taken. Optimum decision metrics and likelihood tests are developed, together with approximated metrics reducing the complexity of the detection block. Moreover, it is illustrated how the soft images produced by the RX nodes may be effectively exploited by tracking algorithms relying on Bayesian filters.

Construction of Near-Optimum Burst Erasure Correcting Low-Density Parity-Check Codes

In this paper, a simple and effective tool for the design of low-density parity-check (LDPC) codes for iterative correction of bursts of erasures is presented. The design method consists of starting from the parity-check matrix of an LDPC code and developing an optimized parity-check matrix, with the same performance over the memoryless erasure channel, and suitable also for the iterative correction of single erasure bursts. The parity-check matrix optimization is performed by an algorithm called pivot searching and swapping (PSS) algorithm. It executes permutations of carefully chosen columns of the parity-check matrix, after a local analysis of particular variable nodes called stopping set pivots. This algorithm can be in principle applied to any LDPC code. If the input parity-check matrix is designed to achieve a good performance over the memoryless erasure channel, then the code obtained after the application of the algorithm provides a good joint correction of independent erasures and single erasure bursts. Numerical results are provided in order to show the algorithm effectiveness when applied to different categories of LDPC codes.

Low-Complexity LDPC Codes with Near-Optimum Performance over the BEC

Recent works showed how low-density parity-check (LDPC) erasure correcting codes, under maximum likelihood (ML) decoding, are capable of tightly approaching the performance of an ideal maximum-distance-separable code on the binary erasure channel. Such result is achievable down to low error rates, even for small and moderate block sizes, while keeping the decoding complexity low, thanks to a class of decoding algorithms which exploits the sparseness of the parity-check matrix to reduce the complexity of Gaussian elimination (GE). In this paper the main concepts underlying ML decoding of LDPC codes are recalled. A performance analysis among various LDPC code classes is then carried out, including a comparison with fixed-rate Raptor codes. The results confirm that a judicious LDPC code design allows achieving achieving a near-optimum performance on the erasure channel, with very low error floors. Furthermore, it is shown that LDPC and Raptor codes, under ML decoding, provide almost identical performance in terms of decoding failure probability vs. overhead.

Pivoting Algorithms for Maximum Likelihood Decoding of LDPC Codes over Erasure Channels

This paper investigates efficient maximum-likelihood (ML) decoding algorithms for low-density parity-check (LDPC) codes over erasure channels. In particular, enhancements to a previously proposed structured Gaussian elimination approach are presented. The improvements are achieved by developing a set of algorithms, here referred to as pivoting algorithms, aiming to limit the average number of reference variables (or pivots) from which the erased symbols can be recovered. Four pivoting algorithms are compared, which exhibit different trade-offs between the complexity of the pivoting phase and the average number of pivots. Numerical results on the performance of LDPC codes under ML erasure decoding complete the analysis, confirming that a near-optimum performance can be obtained with an affordable decoding complexity, up to very high data rates. For example, for one of the presented algorithms, a software implementation has been developed, which is capable to provide data rates above 1.5 Gbps on a commercial computing platform.

Doubly-Generalized LDPC Codes: Stability Bound Over the BEC

The iterative decoding threshold of low-density parity-check (LDPC) codes over the binary erasure channel (BEC) fulfills an upper bound depending only on the variable and check nodes with minimum distance 2. This bound is a consequence of the stability condition, and is here referred to as stability bound. In this paper, a stability bound over the BEC is developed for doubly-generalized LDPC codes, where variable and check nodes can be generic linear block codes, assuming maximum a posteriori erasure correction at each node. It is proved that also in this generalized context the bound depends only on the variable and check component codes with minimum distance 2. A condition is also developed, namely, the derivative matching condition, under which the bound is achieved with equality. The stability bound leads to consider single parity-check codes used as variable nodes as an appealing option to overcome common problems created by generalized check nodes.

On the growth rate of the weight distribution of irregular doubly-generalized LDPC codes

In this paper, the asymptotic growth rate of the weight distribution of irregular doubly generalized LDPC (D-GLDPC) codes is derived. The analysis yields a compact expression which accurately approximates the growth rate function for the case of small linear-weight codewords. This paper generalizes existing results for LDPC and generalized LDPC (GLDPC) codes. Ensembles with smallest check or variable node minimum distance greater than 2 are shown to have good growth-rate behavior, while for other ensembles a fundamental parameter is identified which discriminates between an asymptotically small and an asymptotically large expected number of small linear-weight codewords. Also, in the latter case it is shown that the growth rate depends only on the check and variable nodes with minimum distance 2. An important connection between this new result and the stability condition of D-GLDPC codes over the BEC is highlighted. Such a connection, previously observed for LDPC and GLDPC codes, is now extended to the case of D-GLDPC codes. Finally, it is shown that the analysis may be extended to include the growth rate of the stopping set size distribution of irregular D-GLDPC codes.

On Construction of Moderate-Length LDPC Codes over Correlated Erasure Channels

The design of moderate-length erasure correcting low-density parity-check (LDPC) codes over correlated erasure channels is considered. Although the asymptotic LDPC code design remains the same as for a memoryless erasure channel, robustness to the channel correlation shall be guaranteed for the finite length LDPC code. This further requirement is of great importance in several wireless communication scenarios where packet erasure correcting codes represent a simple countermeasure for correlated fade events (e.g., in mobile wireless broadcasting services) and where the channel coherence time is often comparable with the code length. In this paper, the maximum tolerable erasure burst length (MTBL) is adopted as a simple metric for measuring the code robustness to the channel correlation. Correspondingly, a further step in the code construction is suggested, consisting of improving the LDPC code MTBL. Numerical results conducted over a Gilbert erasure channel, under both iterative and maximum likelihood decoding, highlight both the importance of the MTBL improvement in the finite-length code construction and the possibility to tightly approach the performance of maximum distance separable codes.

On the Growth Rate of Irregular GLDPC Codes Weight Distribution

In this paper the exponential growth rate of irregular generalized low-density parity-check (GLDPC) codes weight distribution is considered. Specifically, the Taylor series of the growth rate is expanded to the first order with the purpose of studying its behavior in correspondence with the small weight codewords. It is proved that the linear term of the Taylor series, and then the expected number of small linear-sized weight codewords of a randomly chosen GLDPC code in the irregular ensemble, is dominated by the degree-2 variable nodes and by the check nodes with minimum distance 2. A parameter is introduced, only depending on such variable and check nodes, discriminating between an exponentially small and exponentially large expected number of small weight codewords.

On a class of doubly-generalized LDPC codes with single parity-check variable nodes

A class of doubly-generalized low-density parity-check (D-GLDPC) codes, where single parity-check (SPC) codes are used as variable nodes (VNs), is investigated. An expression for the growth rate of the weight distribution of any D-GLDPC ensemble with a uniform check node (CN) set is presented at first, together with an analytical technique for its efficient evaluation. These tools are then used for detailed analysis of a case study, namely, a rate-1/2 D-GLDPC ensemble where all the CNs are (7, 4) Hamming codes and all the VNs are length-7 SPC codes. It is illustrated how the VN representations can heavily affect the code properties and how different VN representations can be combined within the same graph to enhance some of the code parameters. The analysis is conducted over the binary erasure channel. Interesting features of the new codes include the capability of achieving a good compromise between waterfall and error floor performance while preserving graphical regularity, and values of threshold outperforming LDPC counterparts.