Control theory approach for implementation of dynamic pricing in mobile cellular networks

Accurate forecasting of future demand for cellular services is essential, as, due to the high infrastructure implementation costs involved, overestimation of demand could be very costly. In addition, the difference between peak and off-peak demand for wireless services can be very significant, both temporally and spatially, and gearing the network to meet peak demand would result in under-utilised network capacity most of the time. It has been suggested that real-time or dynamic pricing (the variation of tariff according to system utilisation) could provide an additional strategy for encouraging more efficient use of available resources. The article tests the effectiveness of a linear dynamic pricing function for QoS and revenue management. An algorithm for determining the optimal linear pricing function, taking into account the service operator's preference for revenue attainment or optimal level of QoS, is suggested; simulation results are derived using OPNET. These show that the linear dynamic pricing function, despite being intuitively easy to understand and implement, is not effective in the control of a complex cellular system. The attained revenue is consistently below the desired level, while call blocking is above the set limit. The results are independent of the assumed elasticity of demand. It is concluded that the assumption of linearity for the dynamic pricing function is the main factor for its constant underperformance and it is recommended that this assumption is dropped in the future.