Simulation of radiographic image data transfer and video conferencing over local area networks

Abstract

One of the objectives of any network installation is to retrieve files from file servers with minimum possible delay. Within three LAN configurations considered in this study, a variable number of servers was considered to vary the service, rate of the server pool. The arrival rate was varied to simulate the number of users and their frequency of requests. A random number of requests for magnified images were considered to articulate the individual choices and requirements. The LAN transmission capacities considered were 1 and 10 Mbps. The objective was then to find the delay for different combinations of these variables. The possibility of video conferencing was also studied along with the radiographic image transfer. The objective of this work was to develop a more generalized simulation model for packetized data transfer over LAN. The possiblity of video conferencing was studied in the Ethernet LAN with local file storage system assuming a probability of 0.01 that a requester may initiate a video conference after studying the magnified images obtained from the file servers. The Ethernet LAN with local storage system gives the lowest delay since each user may the magnified images only. The FDDI LAN works better than the Ethernet LAN with remote file servers because the packet sizes of the FDDI LAN are larger than that of the Ethernet LANS. The packetization time was assumed to be fixed irrespective of the packet sizes. Video conferencing puts excessive demand on the transmission media. As a result it affects all the traffic very seriously and also it does not guarantee quality video image delivery. This simulation can be effectively used to design a network for local area data transfer. The necessary inputs are: number of users, frequency of requests, size and kind of data transfer, and transmission rate of the media. This simulation will give a rational choice on the number of servers necessary for the LAN to be worked on. Although this simulation model was developed for a conceptualized hospital situation, this can be adapted easily to any other similar situations. The main work involved in this thesis is the implementation of a generalized encoder and decoder for BCH codes. The special feature of the implementation is that it works over all finite fields and their extensions as opposed to earlier implementations in literature involving only finite field extensions of the binary field. These generalized codes are attractive mainly because of their rate advantage. Interest in these codewords mapped into modified M-PSK constellations has been generated because of availability of a new method to make a code rotationally invariant. The purpose of this work is to measure the performance of these codes in an Additive White Gaussian Noise ( AWGN ) channel. The effect of Rayleigh fading on the performance has also been simulated for some codes. The new technique of making the M-PSK modulation scheme rotationally invariant is also presented. Chapter I gives an introduction to finite field algebra. Chapter 11 provides an introduction to the terms that are used in channel coding theory. Readers familiar with the basics in Chapters I and 11 may directly skip to Chapter 111. This chapter deals with the encoding rules and the decoding algorithms for BCH codes from an implementation point of view. Chapter IV is the mainstay of the thesis. It explains rotational invariance, presents the new technique and discusses the system issues involved in making a modulation scheme rotationally invariant. The results of the simulation are shown in Chapter V. Appendix A shows an alternative way to make rotationally invariant block codes in a QAM scheme.