AFRICON 2004 POWER SPECIAL ISSUE - TOWARDS NEXT GENERATION COMMUNICATION SYSTEMS CDMA

Jump to:

  1. Recurrent neural networks for sub-optimal multiuser detection
  2. Low complexity constant modulas based cyclic blind adaptive multiuser detection
  3. Evaluating viterbi decoded reed-solomon block codes on a complex spreaded ds/ssma cdma system
  4. An improved new structure of single parity check product codes
  5. Reduced complexity maximum likelihood decoding of linear block codes
  6. Pruned convolutional codes and viterbi decoding using the levenshtein distance metric applied to assynchronous noisy channels
  7. Levenshtein distance-based coding for synchronous, fixed length decoding windows in the presence of insertions/deletions
  8. Insertion/Deletion correction by using paralell-interconnected viterbi decoders
  9. Employing a measure of sparseness to investigate sparse data compression in AWGN conditions
  10. Propogation measurements and multipath channel modelling for Line-Of-Sight19.5 GHZ
  11. On the K-Factor distribution and diffraction fading for Southern Africa
  12. Predicting the long-term average of the effective earth radius factor for south africa using ground based observations
  13. Predicting the monthly average cumulative distribution of the effective earth radius factor for South Africa
  14. Geometric modelling of a spatially correlated mimmo fading channel
  15. Comparison of ADSL theoretical models with a practical scenario

Recurrent neural networks for sub-optimal multiuser detection by N. Moodley and S.H. Mneney
Abstract: This paper explores the use of recurrent neural networks for sub-optimal detection in code division multiple access systems. Research has shown that detectors based on the Hopfield recurrent neural network suffer from localized optimization. The basic Hopfield model is reviewed and we illustrate its use as a multiuser receiver. We investigate the use of stochastic methods to achieve a global minimum solution. A stochastic Hopfield network that employs a probabilistic firing mechanism is proposed for multiuser detection. The performance of the proposed model is investigated via simulation with respect to common linear detectors.
Download Paper
 

Low complexity constant modulas based cyclic blind adaptive multiuser detection by J. B. Whitehead and F. Takawira
Abstract: Periodically time varying (PTV) alters can offer signifcant performance gains over conventional altering methods in DS-CDMA communications systems that are either multirate or corrupted by narrow band interference. This paper develops a new blind adaptive PTV linear multiuser detector based on the frequency shift (FRESH) architecture. The blind adaptive algorithm is based on the linearly constrained constant modulus algorithm (LCCMA), but is modi ed for the FRESH architecture. The resulting stochastic gradient algorithm has signi cantly less computational complexity than the recently proposed recursive and cyclic subspace tracking algorithms, and can offer comparable performance. The robustness of the FRESH-LCCMA algorithm is ensured through a proof of convergence which is carried out in the paper.
Download Paper

Evaluating viterbi decoded reed-solomon block codes on a complex spreaded ds/ssma cdma system by L. Staphorst and L.P. Linde
Abstract: In recent years, the Viterbi Algorithm (VA) has been shown to be an efficient trellis decoder for binary and non-binary linear blocks. In this simulation study the performance of VA decoded non-binary Reed-Solomon (RS) block codes, employed in a wideband Code Division Multiple Access (CDMA) system, operating in multipath fading channel conditions, is considered. The RAKE receiver-based Direct Sequence Spread Spectrum Multiple Access (DS/SSMA) Quadrature Phase Shift Keying (QPSK) system employed in this study uses Complex Spreading Sequence (CSS), such as Zadoff-Chu (ZC), Double Sideband (DSB) Constant Envelope Linearly Interpolated Root-of-Unity (CE-LI-RU) filtered General Chirp-like (GCL), Analytical Bandlimited Complex (ABC) and Quadriphase (QPH) sequences. This paper firstly describes the notion of VA decoding of linear block codes. Background information is then given on the filtered and unfiltered CSS families employed in the simulation study. Next, the complex Direct Sequence Spread Spectrum Multiple Access (DS/SSMA) Quadrature Phase Shift Keying (QPSK) transmitter and RAKE receiver simulator structures, as well as a novel complex multipath fading channel simulator structure are presented. This is then followed by a description of this study’s versatile multi-user multipath fading performance evaluation platform, which is constructed using the complex RAKE receiver-based DS/SSMA QPSK communication system and complex multipath fading channel simulator structures. Lastly, simulated Bit Error Rate (BER) performances are presented for uncoded wideband systems and wideband systems employing VA decoded non-binary RS block codes in realistic multi-user multipath fading channel conditions. The effect of Multi-User Interference (MUI), as well as that of Channel State Information (CSI) on the decoding performance of the VA are also considered.
Download Paper

An improved new structure of single parity check product codes by H. Xu and F. Takawira
Abstract: This paper proposes an improved new structure of Single Parity Check (SPC) product codes with an embedded interleaver. The complexity of decoding the new SPC product codes is the same as the one of the existing single parity check product codes by use of the message passing algorithm. Simulation results and bound analysis demonstrate a significant bite error rate (BER) performance improvement compared to the existing single parity check product codes for high code rates at medium signal-to-noise ratios (SNRs).
Download Paper

Reduced complexity maximum likelihood decoding of linear block codes by S.M. Elengical, F. Takawira and H. Xu
Abstract: This paper proposes a reduced complexity Maximum-Likelihood (ML) decoding Algorithm for Linear Block Codes based on the Kaneko decoder and incorporating ruling out conditions for useless iteration steps. The proposed decoding scheme is evaluated over the Additive White Gaussian Noise (AGWN) channel using Binary Phase Shift Key (BPSK) signalling by simulation. Simulations results show that for negligible performance loss, there is significant reduction in the complexity of decoding.
Download Paper

Pruned convolutional codes and viterbi decoding using the levenshtein distance metric applied to assynchronous noisy channels by L. Cheng and H.C. Ferreira
Abstract: For a convolutional encoding and Viterbi decoding system, two insertion/deletion/substitution (IDS) error correcting techniques are presented in this paper. In the first means, by using the pruned convolutional codes, a rate compatible encoding system can adapt the transmission according to the state of the channel having IDS errors. In the second means, a convolutional encoded sequence is decoded by a modified Viterbi decoding algorithm using the Levenshtein distance metric, in which the IDS errors can be corrected at the same time.
Download Paper

Levenshtein distance-based coding for synchronous, fixed length decoding windows in the presence of insertions/deletions by W.A. Clarke and H.C. Ferreira
Abstract: It is firstly shown that the use of a fixed length decoding window in a channel that introduces insertions/deletions can propagate the errors to such an extent that stronger insertion/deletion correcting codes must be used. It is further shown that random insertion/deletion correcting codes based on the Levenshtein Distance are not optimal for such channels. A modification to the Levensthein Distance is proposed and codes based on this distance are shown to be better suited to a channel with a fixed length decoding window. Rates of the codes based on the Levenshtein Decoding Distance are compared with that of codes based on the Levenshtein Distance.
Download Paper

Insertion/Deletion correction by using paralell-interconnected viterbi decoders by T.G. Swart, H.C. Ferreira and M.P.F. dos Santos
Abstract: A new insertion/deletion correction scheme is presented for standard convolutional codes that makes use of multiple parallel-interconnected Viterbi decoders. Whenever an insertion or deletion error occurs, the connections between different Viterbi decoders ensure that decoding will proceed from the decoder that is in synchronization. In this way, a larger Viterbi decoder is created that can correct insertion and/or deletion errors by extending the Viterbi algorithm to encompass all parallel decoders. Further, it is shown how the performance can be improved by inverting certain bits during the encoding of the convolutional codes. This lowers the frequency of occurrence of repeating sequences, which is detrimental to synchronization when dealing with insertions/deletions.
Download Paper

Employing a measure of sparseness to investigate sparse data compression in AWGN conditions by J. Schoeman and L. P. Linde
Abstract: This paper introduces a measure for data sources that may be used to evaluate the potential compression gain achievable in such sources. The algorithm is applied in two examples, the first being a sparse source coding technique, while the second is the derivation of BER bounds in AWGN for sparse data. The authors present comparative results of sparse source coding and an uncoded system in terms of the introduced performace measure and error propagation in AWGN conditions. Simulation results show that the performance measure can be effectively used to predict the potential of a source for compression and to more accurately estimate the decision interval for detectors, ranging from rd = 0.5493 (Eb/N0 = 0,  = 0.1) to rd = 0.0202 (Eb/N0 = 7,  = 0.4). Comparative PDF and BER results are shown for various sparseness levels in coded and uncoded systems.
Download Paper

Propogation measurements and multipath channel modelling for Line-Of-Sight19.5 GHZ by K. Naicker and S. H. Mneney
Abstract: The need for greater bandwidth and higher data rates will inevitably lead to the usage of the upper portion of the k-band in terrestrial communication systems. The frequency band from 18 GHz to 20 GHz has been identified as a suitable candidate for short-haul line-of sight links. As a consequence, a 6.73 km terrestrial LOS link centered at 19.5 GHz has recently been established between the Howard College and Westville campuses of the University of KwaZulu-Natal (UKZN), with the primary objective to investigate the effect of rain attenuation over the link. Multipath propagation and the resultant frequency selective fading is another major problem afflicting such LOS links. In order to evaluate the effect of multipath propagation over the link statistical models of the channel transfer functions are required. In this paper the research into the simulation and statistical modeling of the multipath propagation for the above-mentioned LOS link will be discussed.
Download Paper

On the K-Factor distribution and diffraction fading for Southern Africa by T.J. Afullo and P.K. Odedina
Abstract: In previous presentations, we have discussed models for determining the probability density function (pdf) of the effective earth radius factor for Botswana as a means of predicting kfactor variation. These models are based on the radiosonde measurement data obtained over three years in Botswana. In this presentation, we compare the k values obtained via the curve-fitting model for Botswana and Durban, South Africa. We compare the median k-factor values and k values exceeded 99.9% of the time. Two design examples show that the type of terrain plays a major role in diffraction fading resulting from k-factor variation. While designing with k=4/3 only results in losses of 2 dB and 5 dB for Durban and Botswana links, respectively, multiple obstructions would result in higher fades. Moreover, the gentler terrain in Botswana results in lower losses that would have been the case if the k-factor for Durban had been 1.1. Also we observe that 99.9% link availability would be harder to attain in Durban than in Botswana, due to the higher k-factor variability and hillier terrain in Durban.
Download Paper

Predicting the long-term average of the effective earth radius factor for south africa using ground based observations by A.J. Palmer and D.C. Baker
Abstract: The work reported in this paper further extends work presented in a companion paper and uses ground based meteorological observations to predict the long-term annual average value of the k-factor. These results are compared with those obtained from the radiosonde data used in the companion paper. The results are in broad agreement with those obtained from radiosonde observations. However, it is found that there are discrepancies in the comparison for Alexander Bay data. This may be due to the influence of the cold Benguela current.
Download Paper

Predicting the monthly average cumulative distribution of the effective earth radius factor for South Africa by A.J. Palmer and D.C. Baker
Abstract: In previous work a model was developed to predict the annual average cumulative distribution of the effective earth radius factor, the k-factor, for telecommunications applications in South Africa. Long term meteorological radiosonde data from eight stations were used for this purpose. In this paper, the model is used to predict the cumulative effective earth radius, or k-factor, distribution for different months as a function of height. Comparisons between observed and predicted results are presented.
Download Paper

Geometric modelling of a spatially correlated mimmo fading channel by B.T. Maharaj and L.P. Linde
Abstract: For a single antenna channel, Shannon’s classical capacity formula predicts a capacity increase of approximately 1 bit/cycle for every 3 dB increase in signal to noise ratio (SNR). Consequently in multiple antenna arrays, with n antenna elements, the increase in capacity can be in the order of the number of antenna elements, n more bits/cycle for each 3dB increase in SNR. This tremendous gain in capacity exhibited by MIMO channels over single antenna systems illustrates the enormous potential in wireless communications systems. This paper presents, for an indoor multielement antenna system, a geometric model incorporating local scatterers around the fixed receiver and transmitter. This model considers uniform scattering at the transmitter and non-isotropic scattering using the von Mises pdf at the receiver. The derived correlation function includes some key elements, such as antenna element spacing, degree of scattering, angle of arrival at user and array configuration in MIMO channels. The capacity of the MIMO fading channel with varying number of antenna elements, the degree of isotropic scattering at receiver, the spacing between the antenna elements and the effect of varying signal to noise ratio are investigated and presented for the chosen model.
Download Paper

Comparison of ADSL theoretical models with a practical scenario by J.H. vanWyk and L.P. Li
Abstract: Asymmetric Digital Subscriber Line (ADSL) is becoming a relatively matured technology as many telecommunication service providers offer this technology to customers. The actual throughput of ADSL is highly dependent on the number of simultaneous users within a cable grouping. In order to guarantee throughput to any user at any time, one must be able to predict the circumstances under which throughput cannot be guaranteed any more. To guarantee high reliability of these predictions, a reliable theoretical model of the system is required. In this paper a comparison is made between the theoretical model and the practical scenario to determine the accuracy of the theoretical model. It was found that the theoretical model closely resembles the practical scenario, which makes QoS estimation possible using a theoretical model, rather than a trial-and-error method.
Download Paper