| dc.description.abstract | Topicality of the problem. GSM mobile networks are very widely used in the world and the number of GSM users exceeds 10^9. In the last decade, the GSM/GPRS network has been very rapidly expanded in Lithuania as well. Coverage of the network has reached 99 % of territory of the country. Load of the network in largest cities reaches up to several hundred Erlangs for square kilometre. Voice traffic still generates the largest part of the network load, however mobile data traffic also grows rapidly. Radio interface between the user of mobile services and the network is not stationary. The received QoS of every user in different places is different. Even in a single cell, radio link characteristics differ in different parts of the cell. The signal level may easily vary about 10–20 dB just within several or tens of meters. Moreover, a mobile user experiences different link conditions while moving in the natural environment within or between cells. However, historically characteristics of QoS of mobile services are measured and evaluated like the ones of the fixed network using statistical methods. Statistically calculated characteristics of QoS do not include information about the received quality of mobile telecommunication services of every individual user. Conception of the received quality of services of individual users (individual QoS – iQoS) in the GSM/GPRS network has not been analysed in the world before. QoS should be measured not only at the network level, as it is done today, but also individually in every user terminal. Possible ways of measurement and evaluation of the received voice and data QoS by an individual user need to be investigated. This would increase satisfaction of users, allow implementation of service level agreements for cellular network customers and introduce new value-added services of high benefit to users (such as quality dependent billing). Aim and tasks of the work. The objective of the dissertation is to investigate methods of measurement, evaluation and improvement of iQoS of data and voice users in mobile telecommunication networks. The tasks for achieving this objective: 1. to create and present a conception of individual quality of services (iQoS) of the user in mobile telecommunication networks and offer the method of iQoS measurement; 2. to develop a system adequate simulation model that will evaluate needs of users and predict the necessary network expansion and the QoS an individual user will receive; 5. for improvement of iQoS for mobile data users to investigate the possibility of transmitting users’ data using parallel mobile data channels. Scientific novelty. Quality of the received voice and data services is a widely analysed topic in the world, however analysing methods of QoS in the mobile network are used only at the OSI network level. Interface between the user and the network is not stationary, therefore statistically calculated QoS does not indicate the actual QoS the individual user receives. In the dissertation a new (not analysed before) conception of the received quality of services of individual users in the GSM/GPRS network is proposed. Aggregation of parallel data channels is a quite old technology, however combination of parallel mobile data channels for connecting remote or moving LANs or WLANs has not been analysed yet. In the dissertation the existing technologies for combining data channels are analysed, modelling as well as an experimental investigation are done and suggestions for technology of combining parallel mobile data channels are formulated. The scope of the scientific work. The dissertation includes 5 chapters. In the first introduction chapter relevance, the objective, the main tasks of the work as well as scientific novelty, practical use of the dissertation results, defended statements and structure of the dissertation are presented. In the second chapter individual quality of services received by the user is investigated. The new conception of the received individual quality of services of users in the GSM network is proposed. It is shown, that QoS should be measured not only at the network level, as it is done today, but also individually in every user terminal. In the third chapter a GPRS/EDGE data transmission technology is analysed. A mathematical model of distribution of common voice and data recourses is made. During the modelling common channel utilization, the number of served users, user data throughput as well as blocking probabilities of voice and data users are analyzed. In the fourth chapter possible solutions of connecting remote or moving LANs using a bundle of parallel GPRS, EDGE and UMTS mobile data channels are investigated. Modelling and measurement results are presented there as well. All investigation results are summarised in the fifth conclusive chapter. At the end of the dissertation the list of references is presented. It includes 117 bibliographical sources and 11 publications of the dissertation author. The total scope of the dissertation is 96 pages of text, 11 tables, 51 figures and 128 references. Conclusions 1. Measurement results in the GSM system indicate that the actually received QoS in different places and for different users is different. Statistical methods for monitoring of QoS, that are used in fixed telephony networks, are not appropriate to introduce any level of certainty of the provided mobile telecommunication services to individual users. For evaluation of individual QoS - iQoS in the network it is proposed to use special software modules installed in mobile phones. This evaluation mechanism would enable to provide objective information about the actually received iQoS to the user as well as to the network provider. This would increase certainty and satisfaction of users, allow implementation of service level agreements for cellular network customers and introduce new value-added services of high benefit to users (such as quality dependent billing) 2. From measurement and modelling results it could be noted that a GPRS/EDGE mobile data transmission channel for an individual (not statistical) user is very different from Frame-relay, Ethernet or other data transmission channels. Throughput and other quality characteristics depend not only on specific characteristics of the radio channel (RxLev, C/I end etc.) but also upon network coverage, user location, characteristics of the user terminal as well as the number of data and voice users. An access channel of an individual user in the mobile data network should be considered as a channel with stochastically varying bandwidth. 3. Parallel GPRS, EDGE and UMTS mobile data channels could be effectively used as a quite cheap and reliable backbone for remotely installed local networks and moving or temporally installed WLAN hot spots. It is especially effective financially when the same operator provides WLAN and mobile data services. From investigation results it could be noted, that, in order to guarantee sufficient data transfer rate and high connection reliability, combination of 2–6 mobile data channels from different base stations of one or several mobile network providers could be done. Quality of every bundled channel should be monitored separately. Then, in the case of connection failure, every channel could be reset individually not influencing connectivity of end users. 4. A protocol, suitable for dynamic per packet load balancing of traffic among several mobile data channels (taking into account bandwidth of channels and priority of packets) does not exist now. To fully utilise a bundle of parallel channels with stochastically varying bandwidth a protocol, supporting fast adaptation of QoS parameters to the channels and fast packet switching, should be developed. | eng |
