The effect of redials and retrials on the GSM Network

Thumbnail Image
Journal Title
Journal ISSN
Volume Title
The scarcity of spectral resources limits the number of channels that can be used on a telecommunication network. Consequently, a user requesting a service particularly a call may be blocked because all channels may be occupied. Initial planning and design of GSM network may not necessarily be accurate because all parameters including the traffic which determines the blocking of the network are based on estimations. Such networks may not meet the industry standard especially when it comes to blocking which is a measure of the Grade of Service (GoS) of the network. A user or the system may experience blocking at one time or the other. Both users and the network may attempt to get a connection. These reattempts constitute redials on the part of the user and retrials on the part of the network and they introduce a phenomenon which cannot be neglected in the analysis of a network. All of these retrials and redials are registered as separate calls. As a result, during the busy periods when blocking is observed in a cell, counters register a much larger volume than the effective call attempts. This research seeks to study of the effect of these reattempts on the GSM network. We seek to find how the blocking probability, redial probability, number of redials and the number of retrials affect the GoS of the network. Four analytical models namely Generalized Redial/Retrial Model (GRRM), Uniform Redial Model (URM), were applied to measured load to extract the excess load which comprise of redials and retrials. When the extraction is complete, the actual load that is supposed to be on the network would be obtained after which an efficient and a better network can be designed. The findings in this study revealed that network traffic is inflated with redials and retrials generated by networks subscribers and the network respectively. All the models when applied produced actual loads that were less than the measured loads. GRRM and SHM proved to be better models to use on highly congested networks while URM and IRM proved to be useful for all congestion conditions. It was also observed that an increase in the blocking probability and the redial probability introduced more load onto the network as the redials and retrials increase. The study has also shown that the network blocking probability may not always meet the industry standard and the number of channels that are used for network operations may often be inadequate.  
A thesis submitted to the Department of Electrical and Electronic Engineering, Kwame Nkrumah University of Science and Technology in partial fulfilment of the requirement for the award of Master of Science degree, November-2012