Search

Neural networks (NNs) are attractive for nonlinear impairment compensation applications in communication systems, such as optical fiber nonlinearity, nonlinearity of driving amplifiers, and nonlinearity of semiconductor optical amplifiers. Without prior knowledge of the transmission link or the hardware characteristics, optimal parameters are completely constructed from a data-driven approach by exploring training datasets, once the NN structure is given. On the other hand, due to computational power and energy consumption, especially in high-speed communication systems, the computational complexity of the optimized NN needs to be confined to the hardware, such as FPGA or ASIC without sacrificing its performance improvement. In this paper, two approaches are presented to accommodate the NN-based algorithms for high-speed communication systems. The first approach is to reduce computational complexity of the NN-based nonlinearity compensation algorithms on the basis of weight pruning (WP). WP can significantly reduce the computational complexity, especially because the nonlinear compensation task studied here results in a sparse NN. The authors have studied an enhanced approach of WP by imposing an additional restriction on the selection of non-zero weights on each hidden layer. The second approach is to implement NNs onto a silicon-photonic integrated platform, enabling power efficiency to be further improved without sacrificing the high-speed operation.

The application of line-of-sight (LOS)- multiple-input multiple-output (MIMO) to the back-haul or fronthaul between a macro-cell and small cells is promising from the viewpoint of low-cost deployment in 4G and 5G cellular heterogeneous networks. This paper presents the bit error rate (BER) performance of low-density parity-check (LDPC)-coded single-carrier 2 × 2 LOS-MIMO using frequency domain equalization (FDE) in the 3rd generation partnership project (3GPP) tapped delay line (TDL) channel models. Computer simulation results show that the average BER of less than 10−8 is achieved with the LDPC coding rate of R = 1/2 or lower in TDL-D and E channel models with the number of non-line-of-sight (NLOS) delayed paths of L NLOS = 13 and 14, that suffer from independent Rayleigh fading. Moreover, we show that high quality such as the average BER of 10−8 is achieved for LDPC- coded 2 × 2 LOS-MIMO using FDE when L NLOS is equal to five or fewer even with the high coding rate of 8/9 in the TDL-E channel model.

The authors analyse and investigate the effectiveness of frequency domain equalisation (FDE) for single-carrier (SC) transmission over a line-of-sight multiple-input multiple-output (LOS-MIMO) channel with frequency selective fading. For this purpose, they propose a 2 × 2 LOS-MIMO system model employing SC-FDE techniques together with a data frame structure having a cyclic prefix and postfix (CPP) overhead. The channel capacity of the proposed system is derived taking into account system design parameters, such as symbol rate and roll-off factor, as well as multipath fading channel parameters. An efficient channel estimation method is also proposed for implementing and simulating the proposed system. Signature curves estimated in simulations are consistent with those expected from the theoretical capacity analysis. The simulation results show that the SC-FDE approach to the LOS-MIMO system can, with small CPP overhead, effectively reduce intersymbol and interstream interference.

In the future cellular systems based on heterogeneous networks (HetNets), the application of line-of-sight (LOS)-multiple-input multiple-output (MIMO) to backhaul or fronthaul between a macro cell and small cells is promising from the viewpoint of low-cost deployment. For this purpose, this paper presents the bit error rate (BER) performance of single-carrier 2 ×2 LOS-MIMO using frequency domain equalization (FDE) in the 3rd generation partnership project (3GPP) tapped delay line (TDL) channel models. Computer simulation results show the effectiveness of LOS-MIMO using FDE in the 3GPP TDL based LOS channel models.

Neural networks are attractive for nonlinear impairment compensation applications in communication systems. In this paper, several approaches to reduce computational complexity of the neural network-based algorithms are presented.

Background: Coronary CT angiography (CCTA) has been applied to detect vulnerable plaques having low attenuation of CT value and positive remodeling which have been associated with acute coronary syndrome. The present study was performed to establish a novel approach to qualitative and quantitative analysis of coronary plaque, especially early detection of high-risk plaque by visualizing intra-plaque localization of necrotic core and the evaluation of plaque stabilizing effect of optimal medical therapy (OMT) including anti-PCSK9 antibody. Methods: We developed color mapped CCTA. The vulnerable plaques are identified by red color (CT attenuation: 0-30) and yellow color (CT attenuation: 30-60). The effect of OMT including anti-PCSK9 antibody on the vulnerable plaques in diabetic patients was studied. Results: Color mapped CCTA clearly demonstrated the presence of necrotic core (red in yellow sign) in vulnerable plaque. After 12 months administration of anti-PCSK9 antibody, an improvement of vulnerable plaque e.g., disappearance of necrotic core in vulnerable plaque was observed. Conclusion: Color mapped CCTA provides a new insight into early detection of high-risk plaque by visualizing intra-plaque localization of necrotic core, also contributes to the evaluation of drugs which may improve vulnerable plaques. Disclosure A. Hirai: None. S. Kondo: None. T. Goto: None. K. Hirai: None. N. Kamiya: None. Y. Tanabe: None. I. Komesu: None. H. Moromizato: None. J. Nakazato: None. S. Ikejima: None. H. Sunagawa: None. T. Kido: None.

We present a novel Wiener smoothing filter procedure for pilot-symbol-assisted estimation of oscillator phase noise. The procedure consists of forward and backward first-order infinite impulse response filtering the pilot symbols and combining the filter outputs to generate an estimate of the phase noise. We analyze the mean-square-error performance of this pilot-symbol-assisted forward–backward Wiener smoothing filter (FBWSF) procedure and show its optimality under the assumption that the phase noise is modeled as a first-order autoregressive moving average process. We then apply the FBWSF as a code-aided phase noise compensation technique in an iterative receiver framework. The iterative receiver presented here first estimates phase noise roughly by using the pilot-symbol-assisted FBWSF; then, it refines the estimates with the code-aided version of FBWSF. The resulting receiver achieves excellent error performance at a feasible level of computational complexity.

This paper presents the bit error rate (BER) performance of frequency domain equalization (FDE) using cyclic-shifted code division multiplexing (CDM) pilot signals for single-carrier line-of- sight (LOS) - multiple-input multiple-output (MIMO) multiplexing. We use different cyclic-shift resources of the same Zadoff-Chu sequence for transmission-stream-specific pilot signals that are essential for estimating the channel response for FDE and phase noise in LOS-MIMO. Computer simulation results show that the required received signal-to-noise power ratio satisfying the BER of 10^(-4) by averaging approximately 4 channel responses after despreading the pilot signal in the frequency domain is decreased by approximately 2.0 dB compared to that without averaging of the channel response for 256QAM, 1024QAM, and 4096QAM in 2x2 and 4x4 LOS-MIMO. Through the simulation results, we show that CDM-based pilot signal multiplexing with different cyclic shifts and the frequency domain averaging of the estimated channel response are effective in achieving accurate channel estimation with a short delay and a low pilot-overhead loss for single-carrier LOS- MIMO.

This paper proposes a symbol timing detection method based on the measured cross-correlation power, i.e., power delay profile, of a pilot sequence for frequency domain equalization (FDE) in single-carrier line-of-sight (LOS) - multiple-input multiple-output (MIMO) transmission. In the proposed method, we detect the best symbol timing that provides the maximum cross-correlation power between the received signal being oversampled with an $N$ times fast Fourier transform (FFT) sampling rate and the pilot sequence replica in the initial acquisition mode. Subsequently, in the tracking mode, we detect the best symbol timing from the maximum cross-correlation power of the pilot sequence at the Lagrange interpolation filter output in a frequency-selective fading channel. Based on computer simulation, we optimize the averaging time of the measured cross-correlation power and the update interval that are used to update the symbol timing in the interpolation filter with the double sampling rate of the FFT sample rate. Computer simulation results show that the proposed symbol timing detection method achieves almost the same bit error rate performance as that for the phase-locked loop employing Gardner's timing error detection for FDE in single-carrier LOS-MIMO in Rummler fading channels.

We study a class of MACs, which we call corruption detectable MAC, that is able to not only check the integrity of the whole message, but also detect a part of the message that is corrupted. It can be seen as an application of the classical Combinatorial Group Testing (CGT) to message authentication. However, previous work on this application has an inherent limitation in its communication cost. We present a novel approach to combine CGT and a class of linear MACs (XOR-MAC) that breaks this limit. Our proposal, \(\textsf {XOR}\text {-}\textsf {GTM} \), has a significantly smaller communication cost than any of the previous corruption detectable MACs, while keeping the same corruption detection capability. Our numerical examples for storage application show a reduction of communication by a factor of around 15 to 70 compared with previous schemes. \(\textsf {XOR}\text {-}\textsf {GTM} \) is parallelizable and is as efficient as standard MACs. We prove that \(\textsf {XOR}\text {-}\textsf {GTM} \) is provably secure under the standard cryptographic assumptions.

Recently, we proposed a high-order circular quadrature amplitude modulation (QAM) constellation scheme with a high low-density parity-check (LDPC) coding rate that is suitable for partial LDPC coding associated with parallel double Gray mapping for microwave mobile backhaul. This paper presents the bit error rate (BER) performance of circular QAM when using a combination of Wiener filter based pilot symbol assisted (PSA) and phase locked loop (PLL) phase noise (PN) estimation and compensation methods. Computer simulation results show that the required received signal-to-noise power ratio (SNR) at the BER of 10−6 using the PSA and PLL PN estimation and compensation is decreased by approximately 0.3, 0.6, and 1.3 dB for the pilot symbol insertion interval of 25, 50, and 100 symbols, respectively, compared to that using only the PSA PN estimation for circular 1024QAM. Accordingly, we show that the PLL PN estimation is effective in decreasing the pilot symbol overhead of PSA PN estimation for circular QAM, which is robust against time-varying PN.

This paper proposes a design for a high-order circular quadrature amplitude modulation (QAM) constellation using partial low-density parity-check (LDPC) coding associated with parallel double Gray mapping aiming at the application to a microwave radio backhaul to deal with time-varying phase noise channels. In the proposed circular QAM constellation design, we first design the best constellation in which the same number of signal points is mapped to all concentric rings. Then, we decrease the number of signal points in the inner rings and remap the signal points to the newly added rings outside the original rings while maintaining high affinity to the double Gray mapping feature. Computer simulation results show that partial LDPC coding with double Gray mapping decreases the required received signal-to-noise power ratio (SNR) at the bit error rate (BER) of 10−6 by approximately 0.2 dB compared to full LDPC coding for a high LDPC coding rate such as 0.9. We also show that the proposed circular 1024QAM decreases the required received SNR at the average BER of 10−6 by approximately 0.5 dB compared to that for rectangular 1024QAM.

In this paper, we present a novel phase error estimation scheme for low-density parity-check (LDPC) coded high-order quadrature amplitude modulation (QAM) transmission systems, and we provide a theoretical analysis of its performance. The scheme uses known pilot symbols which are assumed to be placed at equally spaced positions, and phase errors occurred at these pilot symbol positions are first estimated. They are then interpolated over the other positions by using a Wiener filter (WF), and estimation accuracy is next refined by using an LDPC code-aided phase-locked loop (PLL). We present an efficient implementation of this two-stage, Pilot-Symbol-Assisted and Code-Aided (PSA/CA) scheme and evaluate its performance by means of power spectrum analysis. We determine the mean square error at each step of the PSA/CA scheme, and we find an optimal value of the noise equivalent loop bandwidth of the PLL under the assumption that local oscillator phase noise can be modeled as a single-pole single-zero model. We show through simulations that the PSA/CA scheme has a very high phase noise tolerance comparable to that of state-of-the-art phase estimation schemes and provides a computationally efficient alternative.

This paper shows that several attractive classes of quasi-cyclic (QC) low-density parity-check (LDPC) codes can be obtained from affine planes over finite fields. One class of these consists of duals of one-generator QC codes. Here, for codes contained in this class, the exact minimum-distance and a lower bound on the multiplicity of the minimum-weight codewords are presented. It is shown that the lower bound on the multiplicity provides a very accurate indication of the bit error performance at moderate and high signal-to-noise ratios, and thus error-floor behavior can be easily predicted. Also discussed is a class consisting of codes from circulant permutation matrices. An explicit formula for the rank of the parity-check matrix is presented for these codes. Furthermore, it is shown that each of these codes can be identified as a code constructed from a constacyclic maximum distance separable (MDS) code in a similar manner to the RS-based LDPC codes presented by Chen et al. and Djurdjevic et al. Experimental results show that a number of high rate QC-LDPC codes with excellent error performance are contained in these classes

Finite geometry codes are defined as the null spaces of the incidence matrices of points and flats in finite geometries. In this paper, we investigate the incidence matrix of points other than the origin and lines not passing through the origin in the affine plane AG(2,2s), and we present two classes of quasi-cyclic codes derived from submatrices of the point-line incidence matrix. We also investigate the 2-ranks of those submatrices.

Three algebraic soft-decision decoding algorithms are presented for binary Bose-Chaudhuri-Hocquengham (BCH) codes. Two of these algorithms are based on the bounded distance (BD)+1 generalized minimum-distance (GMD) decoding presented by Berlekamp (1984), and the other is based on Chase (1972) decoding. A simple algebraic algorithm is first introduced, and it forms a common basis for the decoding algorithms presented. Next, efficient BD+1 GMD and BD+2 GMD decoding algorithms are presented. It is shown that, for binary BCH codes with odd designed-minimum-distance d and length n, both the BD+1 GMD and the BD+2 GMD decoding algorithms can be performed with complexity O(nd). The error performance of these decoding algorithms is shown to be significantly superior to that of conventional GMD decoding by computer simulation. Finally, an efficient algorithm is presented for Chase decoding of binary BCH codes. Like a one-pass GMD decoding algorithm, this algorithm produces all necessary error-locator polynomials for Chase decoding in one run.

This letter presents a class of quasi-cyclic (QC) low-density parity-check (LDPC) codes which are efficiently encodable and have better memory efficiency than other randomly constructed QC-LDPC codes. The parity-check matrix of the codes is obtained from a block circulant matrix that is generated from a self-reciprocal generator polynomial of a maximum distance separable (MDS) code by replacing some of its block entries by all zero matrices. We show by simulations that the codes can achieve excellent error performance.

We describe an efficient algorithm for successive errors-and-erasures decoding of BCH codes. The decoding algorithm consists of finding all necessary error locator polynomials and errata evaluator polynomials, choosing the most appropriate error locator polynomial and errata evaluator polynomial, using these two polynomials to compute a candidate codeword for the decoder output, and testing the candidate for optimality via an originally developed acceptance criterion. Even in the most stringent case possible, the acceptance criterion is only a little more stringent than Forney's (1966) criterion for generalised minimum distance (GMD) decoding. We present simulation results on the error performance of our decoding algorithm for binary antipodal signals over an AWGN channel and a Rayleigh fading channel. The number of calculations of elements in a finite field that are required by our algorithm is only slightly greater than that required by hard-decision decoding, while the error performance is almost as good as that achieved with GMD decoding. The presented algorithm is also applicable to efficient decoding of product RS codes.

We propose and investigate a novel FEC coding scheme based on a concatenation of QC-LDPC with single-parity-check codes. High performance of a Q-limit of 5.8 dB with 20.5% overhead has been achieved with FPGA-based simulations.

In 1981, Goppa established an epoch-making method for constructing a code utilizing the theory of algebraic curves [1]. This is called the algebraic curve code. This paper develops the theory on an arbitrary curve, which may contain singular points and discusses the method of constructing the code. The method employed in this paper is to eliminate the singular point by applying the transformation called blowing-up and to construct the code on the nonsingular model. The number of genes of curves can be derived simultaneously. It is made easy to derive the bases of the linear spaces L(G) and Ω(G-D), as in the case of the nonsingular curve. The algebraic curve C considered in this paper is the irreducible projected plane curve or the algebraic curve which is obtained by gluing irreducible affine plane curves. In this case, the nonsingular model of C can also be represented as a gluing of irreducible affine plane curves. Then the method is applied to the hyperelliptic plane curve on which the number of rational points achieves the Hasse-Weil bound or certain hypernonelliptic plane curves. It is shown that the precise structure of the code can be examined.

In this paper, we present an efficient systematic encoding algorithm for quasi-cyclic (QC) low-density parity check (LDPC) codes that are related to cyclic maximum-distance separable (MDS) codes. The algorithm offers linear time complexity, and it can be easily implemented by using polynomial multiplication and division circuits. We show that the division polynomials can be completely characterized by their zeros and that the sum of the respective numbers of their zeros is equal to the parity-length of the codes.

We propose a fast decoding algorithm for a class of geometric Goppa codes defined on certain algebraic plane curves, associated with Artin-Schreier extensions of F/sub q/(x), introduced by Stichtenoth [3]. Although we do not attempt here to treat all the class of curves introduced by Stichtenoth, we do include certain elliptic, hyperelliptic and Hermitian curves. These curves are defined by the homogeneous equation Y/sup a/Z/sup b-a/ + YZ/sup b-1/ = X/sup b/ over an arbitrary finite field F/sub q/ of characteristic p, where a and b are relatively prime integers such that a = P/sup v/(V /spl epsiv/ N/sup */), a < b and the zeros of Y/sup a/ + y form an additive subgroup of F/sub q/ of order p/sup v/. The main step of the proposed algorithm is to solve a key equation studied by Porter, Shen and Pellikaan [1]. For this purpose, we derive explicit formulas for certain differential forms, which are used to construct the syndrome of the codes defined on the above-mentioned curves, and propose a modified version of Sakata's algorithm [4]. Further, we prove, in work inspired by Shen's study [2], that the Porter-Shen-Pellikaan key equation for codes defied on the curves treated here can be solved by using our modified Sakata algorithm with complexity O(d/sup 2//sub des/a + g/sup 2/a), where d/sub des/ is the designed minimum distance and g is the genus of the curve. The proposed decoding algorithm may be regarded as an extension of Shen's algorithm [2] for Hermitian codes to a wider class of codes. For certain hyperelliptic codes, this algorithm can decode up to [(d/sub des/ - 1)/2] errors with complexity O(n/sup 2/), where n is the word length of the code.

In this paper, we investigate the properties of the point-line incidence matrix of a finite affine plane and present a construction of quasicyclic LDPC codes based on submatrices of the incidence matrix. The codes from the point-line incidence matrices of the affine and projective geometries perform very well with iterative decoding algorithms for low-density parity-check (LDPC) codes.