An optimized neural network for monitoring Key Performance Indicators in HSDPA

HSDPA (High Speed Downlink Packet Access) is drawing great attention as the 3.5G technology capable of providing higher data rate packet switch services over Universal Mobile Telecommunication System (UMTS) to support broadband services like multimedia conferencing, VoIP, or high-speed internet access. The paper proposes the use of a Learning Vector Quantization (LVQ) Neural Network able to estimate the quality of service (QoS) across analysis of Key Performance Indicators (KPIs) and to provide automatically a possible classification of warnings related to the load status of HSDPA radio resources or to the bad radio channel quality condition.