Andrei Alexandru Enescu

On the influence of the forgetting factor of the RLS adaptive filter in system identification

The overall performance of the recursive least-squares (RLS) algorithm is governed by the forgetting factor. The value of this parameter leads to a compromise between low misadjustment and stability on the one hand, and fast convergence rate and tracking on the other hand. In this paper, we analyze another important phenomenon that has to be considered when choosing the value of the forgetting factor. Considering a system identification setup, there is a “leakage” of the system noise into the output of the adaptive filter. This process is highly influenced by the value of the forgetting factor but it also depends on the length of the adaptive filter. Simulations performed in an echo cancellation configuration prove these theoretical findings.

Long Term Evolution primary synchronization algorithms

In this paper we will describe the procedure for primary synchronization in the Long Term Evolution (LTE) systems downlink and we will propose two enhanced algorithms that rely on the traditional methods described in other papers (based on cross-correlation properties of Zadoff-Chu sequences), but provide extra robustness through a validation process that can be made either “hard” or “soft”. The result contains the value for the sector identifier and location of the primary sequence, necessary for further synchronization and confirmation through the cell-specific reference signal (RS).

CTC Turbo Decoding Architecture for H-ARQ Capable WiMAX Systems Implemented on FPGA

This paper presents the results of a hybrid automatic request (H-ARQ) capable, convolutional turbo decoder implementation on field programmable gate array (FPGA). It emphasizes area, speed, and the decoding performances when using fixed point implementation. The convolutional turbo coding scheme is the one specified in the 802.16e-2005 standard. The code data rates are 1/2, 2/3, and 3/4, the maximum number of iterations is 8, and the block size is from 6 to 600 bytes. The maximum number of retransmissions for H-ARQ block is 4, according to the mandatory profiles of the 4G communication systems.

Super-orthogonal space-time trellis codes with eight dimensional phase-shift keying signal constellations

We propose a new method of designing super-orthogonal space-time trellis codes (SOSTTCS) by labeling each state transition of the trellis diagram with an ordered pair of 2×2 orthogonal matrices instead of only a single 2×2 orthogonal matrix. An 8D instead of a 4D signal constellation is thus used for modulation and the delay elements of the trellis encoder are 4T instead of 2T, where T is the 2D symbol duration. Therefore, at the receiver, the sequence decoding using Viterbi algorithm requires two times less frequent updating of the branch metrics. We investigate by computer simulation the performance of the new sosttcs compared to sosttcs designed by the known method. The frame error rate (fer) and the bit error rate (ber) simulation results show that the performance of the new sosttcs is practically as good as that of the sosttcs of similar complexity designed along classical lines, with the advantage that the number of arithmetic and logic operations performed by the decoder per time unit is smaller, which is crucial in high data-rate applications.RésuméNous proposons une nouvelle méthode pour concevoir des codes espace-temps superorthogonaux (sosttcs) en étiquetant chaque transition du diagramme en treillis par une paire ordonnée de matrices orthogonales 2×2 à la place d’une seule matrice orthogonale 2×2. On emploie donc pour la modulation une constellation de signaux 8D à la place d’une 4D, ce qui fait que les éléments de retard sont 4T à la place de 2T, où T est la durée d’un symbole. Pour cette raison, au récepteur, le décodage séquentiel utilisant l’algorithme de Viterbi demande une actualisation deux fois moins fréquente des métriques de branche. Par simulation au moyen d’un ordinateur, nous étudions la performance de nos nouveaux codes en treillis espace-temps super-orthogonaux en les comparant aux codes conçus par la méthode connue. Les résultats des simulations concernant le taux d’erreurs de trames et le taux d’erreurs de bits (teb) montrent que la performance de nos nouveaux codes espace-temps super-orthogonaux est, d’un point de vue pratique, aussi bonne que celle des codes de complexité similaire conçus de manière classique, avec l’avantage d’un nombre réduit d’opérations arithmétiques et logiques à effectuer par le décodeur par unité de temps, ce qui compte le plus dans les applications à débit élevé.

A network echo canceler based on a SRF QRD-LSL adaptive algorithm implemented on Motorola StarCore SC140 DSP

A novel network echo canceler based on a square root free QR-decomposition-based least-squares lattice (SRF QRD-LSL) adaptive algorithm is presented. Specific characteristics of this type of algorithms are their high convergence rate, numerical stability and superior robustness to double-talk, which is not present in the case of the normalized least mean squares (NLMS) family of algorithms. This echo canceler is implemented on a powerful fixed-point digital signal processor (DSP), Motorola SC140. The experimental results prove superior performances over the classical echo cancelers based on NLMS algorithms.

Recursive Least-Squares Lattice Adaptive Algorithm Suitable for Fixed-Point Implementation

This paper proposes a version of recursive least-squares lattice (RLSL) adaptive algorithm suitable for fixed-point implementation. It is based on a modified form of the cost function in order to obtain an asymptotically unbiased estimator of the mean-square error, reducing significantly the dynamic range of the algorithms parameters. Moreover, in order to achieve an enhanced numerical accuracy an error-feedback mechanism is employed. These improvements lead to facilities for fixed-point implementations.

Sub-optimal solutions to code synthesis for space-time diversity

Transmit diversity is a very attractive solution for efficient communications through fading wireless channels. The paper presents some aspects concerning the synthesis of the coding matrix in a multiantenna configuration. Because no optimal (maximum coding advantage, maximum transmission rate, minimum delay) codes can be found for systems with more than two antennas, some sub-optimal solutions are analyzed.

Gradient Adaptive Lattice Algorithm Suitable for Fixed Point Implementation

The lattice-based implementations of the gradient adaptive algorithms offer improved convergence rate capabilities, especially for high correlated inputs. In the classical gradient adaptive lattice (GAL) algorithm a division operation per stage is used, which significant grows the computational complexity in a fixed-point implementation context. In this paper, an approximate version of the GAL algorithm is proposed. This algorithm replaces the division operation by three multiplication operations and one addition operation. The simulation results prove that this algorithm keeps the rate of convergence and the good numerical properties of the classical GAL algorithm.

An overview of soft-output sphere decoders with constant throughput

This paper represents an overview of the main sphere decoders used in MIMO (Multiple Input Multiple Output) spatial multiplexing communication systems. The family of sphere decoders addressed in this paper exhibits fixed throughput, regardless of the radio channel conditions. The algorithms are compared with respect to their complexity and performances when used with a soft Viterbi channel decoder. The paper concludes with the criteria for choosing the best algorithms in a given configuration and for specific modulation scheme.

Spatial Multiplexing Turbo Receiver with Reduced Complexity

This paper proposes a multi-antenna spatial multiplexing receiver exhibiting very good performances in the presence of fading and low signal-to-noise ratio. These result from the use of a soft-output sphere decoder with reduced complexity and a robust turbo channel decoder affected by log-likelihood ratio quantization and saturation. Fixed-point implementation is targeted, taking in account the receiver throughput and latency.