Bartosz Mielczarek

Scenario-based performance analysis of routing protocols for mobile ad-hoc networks

This study is a comparison of three routing protocols proposed for wireless mobile ad-hoc networks. The protocols are: Destination Sequenced Distance Vector (DSDV), Ad-hoc On demand Distance Vector (AODV) and Dynamic Source Routing (DSR). Extensive simulations are made on a scenario where nodes moves randomly. Results are presented as a function of a novel mobility metric designed to reflect the relative speeds of the nodes in a scenario. Furthermore, three realistic scenarios are introduced to test the protocols in more specialized contexts. In most simulations the reactive protocols (AODV and DSR) performed significantly better than DSDV. At moderate traffic load DSR performed better than AODV for all tested mobility values, while AODV performed better than DSR at higher traffic loads. The latter is caused by the source routes in DSR data packets, which increase the load on the network. routers and hosts, thus a node may forward packets between other nodes as well as run user applications. Mobile ad-hoc networks have been the focus of many recent research and development efforts. Ad-hoc packet radio networks have so far mainly concerned military applications, where a decentralized network configuration is an operative advantage or even a necessity. Networks using ad-hoc configuration concepts can be used in many military applications, ranging from interconnected wireless access points to networks of wireless devices carried by individuals, e.g., digital maps, sensors attached to the body, voice communication, etc. Combinations of wide range and short range ad-hoc networks seek to provide robust, global coverage, even during adverse operating conditions.

Performance of coherent UWB Rake receivers with channel estimators

In this paper, the performances of DS-UWB and TH-UWB for a single-user link are compared for a data rate of 100 Mbps using the newly proposed, realistic UWB channel model from IEEE P802.15. The performance of suboptimal, fractionally spaced (FS), coherent Rake receivers with a pulse-matched filter is numerically evaluated with two channel estimation algorithms. DS-UWB and TH-UWB with antipodal modulation perform basically the same. The FS-Rake receiver performs much better than chip- and symbol-spaced Rake receivers. The best evaluated channel estimation algorithm gives only a fractional loss compared with the perfectly estimated channel. However, since TH-UWB is more difficult to synchronize, the conclusion is that DS-UWB is more suitable for high-speed indoor links.

Phase offset estimation using enhanced turbo decoders

This paper discusses a realistic turbo coding system with the signal phase which has not been perfectly estimated. We propose improved decoding algorithms for the situations when the residual phase error can be modelled by the Gaussian probability distribution and a Markov chain, a model which can be used in many actual phase estimators. It is shown that increasing the state space of the decoders can decrease the bit error probability.

Timing error recovery in turbo-coded systems on AWGN channels

The classical method for decoding turbo codes is the Bahl-Cocke-Jelinek-Raviv (1974) algorithm which, in the majority of existing papers, is analyzed under the assumption of perfect channel knowledge and synchronization. In reality, however, its properties change significantly when synchronization is not ideal. We show that the influence of timing synchronization offset can be modeled as a decreased effective signal-to-noise ratio, which provides results consistent with simulations. Furthermore, we present a general model for the soft bit output of the BCJR algorithm and apply it to the situation when the received signal is sampled with a timing offset. Finally, we use the derived properties of the algorithm in a simple timing synchronizer which is shown to outperform the classical maximum-likelihood non-data-aided symbol synchronizer without a need of using additional pilot symbols.

Adaptive Hybrid ARQ Systems With BCJR Decoding

We propose and evaluate a novel method of constructing Hybrid Automatic Repeat reQuest (HARQ) systems using the specific properties of the Bahl, Cocke, Jelinek, and Raviv (BCJR) error-correcting algorithm. Because the convergence to the actual codeword is not always guaranteed with the BCJR approach, we propose a system in which two different types of Negative AcKnowledgement messages (NAKs) are employed. The first type is the conventional 1-bit NAK, and the second type specifies retransmission pattern in such a way that the additional parity bits are concentrated on the parts of the code trellis that did not converge to a valid sequence. This is different from the traditional construction of rate-compatible punctured codes (RCPCs), which attempts to obtain the optimal weight distance properties of the codes without taking the convergence properties of the BCJR decoder into account. We demonstrate the performance of the algorithm using RCPCs, and we show that our system outperforms the best known conventional HARQ scheme in terms of the throughput and the average length of retransmitted blocks on practical Gaussian, Rayleigh, and thresholded Rayleigh channels. Moreover, as opposed to other adaptive HARQ algorithms, our solution requires no precomputed lookup tables, and it is robust to changes in the channel characteristics and only introduces moderate increase in feedback link throughput requirements.

Vector Quantization of Channel Information in Linear Multi-User MIMO Systems

In this paper, we propose a new vector quantization (VQ) algorithm for encoding channel state information feedback in multiple antenna, multi-user systems operating on flat fading channels with rich scattering. We consider an approach where the receiver chooses an instantaneous throughput maximizing modulation matrix from a finite set of predefined matrices (codewords). The codebook of modulation matrices is constructed based on joint optimization of the dominant channel eigenmodes of users and separate quantization of power levels. The proposed algorithm is very flexible and can be used in a variety of system configurations, including varying number of receiver antennas and frequency selective channels. We implement the proposed algorithm on flat fading MIMO channels and show the influence of the feedback rate on system capacity. We demonstrate that, even with low feedback rate, the ergodic capacity of the proposed system closely approaches the theoretic capacity of the system with perfect channel state information at the transmitter

Flexible channel feedback quantization in multiple antenna systems

We propose a new vector quantization (VQ) algorithm for reducing the bit rate used for channel state information feedback in a variety of multiple antenna systems on flat and frequency selective channels. We consider an approach where the receiver chooses an instantaneous capacity maximizing entry from a small finite set of predefined covariance matrices. The codebook of covariance matrices is constructed based on separate optimization of the channel eigenvectors and power allocation among them. If, for the given channel realization, one of the predefined covariance matrices provides higher capacity than equal power distribution, the indices of the two codebooks are fed to the transmitter; otherwise, the transmitter uses the open-loop approach. We implement the proposed algorithm on flat fading and frequency selective MIMO channels and show the influence of the feedback rate on system capacity. For the case of flat fading, the required feedback rates are approximately equal to the product of the number of transmit and receive antennas. Although much higher feedback throughput is needed for reliable channel state information feedback in OFDM systems, we show a simple clustering technique to lower the required bit rate.

Influence of CSI Feedback Errors on Capacity of Linear Multi-User MIMO Systems

In this paper, we discuss the influence of feedback channel errors on throughput of linear multi-user multiple-input multiple-output (MIMO) systems employing vector quantization (VQ) algorithms for encoding channel state information (CSI). We consider an approach where the receiver estimates the downlink channel and chooses one of the predefined channel characterization codewords whose indices are transmitted back to the transmitter. In practice, the feedback channel will introduce errors in transmission of the indices and we evaluate the resulting channel capacity loss for different system setups. We demonstrate the performance of two types of systems, one using time division multiplexing of individual users and another transmitting to multiple users. We evaluate two basic error mitigation approaches: the first one with user expurgation using error-detection codes and the second one, based on specially designed VQ indexing. We show that the proposed schemes are very robust to even relatively high bit error rates in the feedback link and that proper VQ indexing may suffice for actual system design, even if the feedback channel is not very reliable.

Modeling fading channel-estimation errors in pilot-symbol-assisted systems, with application to turbo codes

In this paper, we address the issue of imperfect channel estimation in coded systems on fading channels. Since performance of channel codes is influenced in different ways by different components of channel-estimation errors, we develop a simplified model which separates the estimation errors of a Wiener-filtered received signal into the amplitude error and the phase error. Based on the model, we derive tight bounds on component error variances. Moreover, we prove that the classical Wiener filter results in a biased estimate of the channel amplitude. We also show that the probability of having a phase-estimation error large enough to cause decision errors in the receiver is significant. Using our model, we derive an approximate upper limit on the optimum pilot-symbol spacing and approximate lower limit on bit-error rate performance of coded systems with a given pilot-symbol separation. The proposed model and derivations are confirmed by extensive simulations.

Improved iterative channel estimation and turbo decoding over flat-fading channels

In this paper, we discuss the influence of imperfect Wiener channel estimation on performance and dynamics of a turbo decoder using nonlinear soft data feedback. We show that the correlation between the channel estimates and soft bits produced in subsequent iterations can lock the turbo decoder in a wrong minimum and create a high error floor. This problem can be circumvented by a different way of estimating code symbols used to improve the synchronization of the signal. An enhanced version of the channel estimator using Wiener filtering of the initial pilot symbols, followed by a modified Viterbi decoder and turbo decoder is shown to improve the systems performance by around 0.5 dB in the waterfall region and to considerably lower the error floor.