On the design, implementation and experimental evaluation of a novel gateway architecture for the GSM Short Message Service

Congestion and capacity pi'ohIens of the existing mobile communication networks of the late eighties resulted in the demand for a brand new mobile telephony standard. Competition amongst the various existing standards implementation was fierce and lead to the availability of a plethora of incompatible networks and very little hope for the establishment of a global technology geared up to the expectancy of the users. Paging network were widely used as a cheaper alternative to voice enabled networks and were Offering users the ability to receive textual information while on the move. Bridging the gap between the paging world and the mobile communication world was essential. The Global System for Mobile communications was designed as the European answer. The European alternative was promising a feature rich, secure and truly global system with the added ability to handle two-way paging like functionalities with the Short Message Service (SMS). The success of a new service such as SMS relies heavily on its adoption by a majority of users. which in turn is mainly a consequence of the availability of software application gateways. In the mid nineties a Sheffield based software company realising the market needs teamed up with the University's research department to produce the first commercially available gateway architecture for the SMS. The resulting work is described in this thesis. The first part of the analysis examines the architecture of' a (ISM network and the building blocks of the SMS. The technical implementation is described and the fundamental properties examined such as roaming. routing, protocol limitations, usability and interoperahility problems. The specification and design of the gateway architecture is then addressed with an emphasis on character set conversion, routing and queLicing issues. The implementation details are then examined with a description of each of' the modules. The performance of the gateway is examined with the implenientation of a test bed fed by traffic generated by customers. The issues examined were: identification of bottlenecks, protocol efficiencies and an analysis of the chosen queueing model implementation. The second part of the analysis presents the results obtained from the measurements taken for a period of a year. Arm analytical model was formulated to validate the results from the measurements. The comparison revealed the ability of the model to simulate the behaviour of the gateway under medium to heavy loads and highlighted the areas that would be most affected by optimisation. The important factors limiting throughput and quality of service were di scoverecl in the capacity of the connections to network operators and policy chosen for the message queues. An alternative queueing discipline is proposed that would lead to increased fairness offered to the wide variety of applications connected to the gateway and the network operators through a single link of' known capacity. Interactive conversations such as quizzes and gaines based are offered low latency while more bandwidth demanding ones such as mass voting applications benefit from very high throughput. The overall gateway,architecture described was the first one of its kind and consequently each of its module was designed and implemented from scratch. As a result Dialogue (Tommnunicatmons benefited l'i'omn the novelty and head start needed in a fierce competitive market to position itself as a market leader ]it mobile applications and services, competing with global companies such as Logica. Ericcson or Microsoft.