Search

Although the wireless communication systems offer limitless possibilities of service utilization, these systems are not error free. Variable error rate of these systems is the main reason of being concerned with error control techniques. Selection of suitable error control techniques is very important and depends on performance. We can choose from three basic techniques Send-and-wait, Go-back-N and Selective repeat. Nowadays, Hybrid Automatic repeat request techniques have the best aptitude for solving the problem with error detection. The thesis deals with division and functioning of Automatic repeat request schemes. In particular, we focus on the efficiency criterions of Automatic repeat request schemes--throughput and computational complexity. These criterions are presented based on simulations for Go-back-N scheme.

We present a new measurement of the spin-dependent structure function $g_{1}^{\rm d}$ of the deuteron from deep inelastic scattering of 190 GeV polarized muons on polarized deuterons. The results are combined with our previous measurements of $g_{1}^{\rm d}$. A perturbative QCD evolution in next-to-leading order is used to compute $g_{1}^{\rm d}(x)$ at a constant $Q^{2}$. At $Q^{2} = 10$ GeV$^{2}$, we obtain a first moment $\Gamma_{1}^{\rm d} = \int_{0}^{1} g_{1}^{\rm d}{\rm d}x = 0.041 \pm 0.008$, a flavour-singlet axial charge of the nucleon $a_{0} = 0.30 \pm 0.08$, and an axial charge of the strange quark $a_{s} = -0.09 \pm 0.03$. Using our earlier determination of $\Gamma_{1}^{\rm p}$, we obtain $\Gamma_1^{\rm p} - \Gamma_1^{\rm n} = 0.183 \pm 0.035$ at $Q^2 = 10\,\mbox{GeV}^2$. This result is in agreement with the Bjorken sum rule which predicts $\Gamma_1^{\rm p} - \Gamma_1^{\rm n} = 0.186 \pm 0.002$ at the same $Q^2$.

SPIHT algorithm is well known reference for image coding algorithms. In this paper we present its specific extension called SM SPIHT related to region of interest encoding (ROI) but extending this approach further, ending with individual weight of importance for each pixel in image using the form of saliency map (SM). With carefully designed saliency map the approach has proven to be effective. The properties of proposed algorithm are presented and discussed in detail and further possible directions are outlined.

Significant advance in the networks design led to the need of optimizing the future communication protocols with the aim to evaluate the network performance. The most common techniques include simulation, analytical models and analysis of empirical data. Accurate error process modelling is necessary for understanding the network behaviour and is essential in designing of error control protocols or real time applications. Discrete Markov models, mainly the Gilbert model, are commonly used for modelling the network error characteristics. This approach is based on the analysis of a communication on link. In this paper we present error model for packet loss and subsequent error description. The proposed model simulates the error performance of wireless communication link. The model is designed as two independent Markov chains, where the first one is used for packet generation and the second one generates correctly and incorrectly transmitted bits for received packets from the first chain. The statistical analyses of real communication on the wireless link are used for determination of model’s parameters. The obtained results generated by proposed model are compared with the real data collection. It is also designed and programmed generator for received traffic. First part of this generator is a generator for packet loss. Second part is a generator for bit error rate. Using the obtained parameters and the implementation of the generator, we collected generated traffic. The obtained results generated by proposed model are compared with the real data collection.

This paper introduces a method for restoration of missing arbitrarily shaped image segments using combined dircrete orthogonal trunsforms (DOT). The missing segment texture is successively interpoIate8/extrapolafed using the texture of the correctly decoded segments belonging to the same object, rhus having the same texture pattern.

This paper presents a new scheme for lossless image coding with SCAN image data scanning (Maniccam et al. (2001)). A methodology for creation of a convenient space-filling curve based on correlation is introduced.

An optimum adaptive go-back-N (GBN) scheme is studied. The considered scheme is an enhancement of recently proposed adaptive schemes in the papers by Yao (1995), Chakraborty and Liinaharja (1999), with principles of the optimum GBN scheme presented by Bruneel and Moeneclaey (1986) in their study. An exact throughput analysis of the scheme is presented.

An efficient description of image segments having an arbitrary shape is one of the main problems in object oriented coding. In our contribution, we introduce a scheme for region based coding derived from the BTC (block truncation code) algorithm (see Farkas, P. et al., "Adaptive Image Coding Based on Generalized IBTC", Broadband Islands 1994: Connecting with the End-User, Elsevier Science B.V., 1994). After segmentation, contours of the created regions are described by an 8-directional chain code as in MPEG-4. In the interior of the regions, there is applied a modified BTC algorithm that we call RTC-region truncation code. The results of compression are compared with BTC. Especially for low bit rate image coding, we achieved, in some cases, a better visual quality using similar values of an objective criterion parameter (MSE). In our approach, it is not necessary to follow the rules for the BTC channel encoder.

A continuous automatic repeat request (ARQ) scheme is considered which is an extension of the classic go-back-N ARQ scheme and a simplification of the optimum go-back-N ARQ strategy proposed by Bruneel and Moeneclaey (1986).