A Cognitive Scheme For Radio Admission 
2012 3rd International Workshop on Cognitive Incromation Processing (CIP)
A Cognitive Scheme for Radio Admission Control in LTE systems
Biljana Bojovic, Nicola Baldo and Paolo Dini
Centre Tecnol`ogic de Telecomunicacions de Catalunya (CTTC), IP Technologies Area
Av. Carl Friedrich Gauss 7, Castelldefels (Barcelona), Spain
{bbojovic, nbaldo, pdini}@cttc.es
Abstract— In order to provide QoS requirements in high speed future communication networks, such as LTE, the operator has to provide a Radio Admission Control algorithm which will guarantee the QoS of different service types (e.g. voice, data, video, ftp) while maximizing radio resource utilization. In this paper we propose a Cognitive Radio Admission Control Scheme based on a Multilayer Feed-forward Neural Network. According to our scheme, the eNodeB performs Radio Admission Control using a cognitive engine that learns from the past experience how the admission of a new session would affect the QoS of all sessions in the future. We implemented our Radio Admission
Control Scheme using a LTE-EPC simulator. Since our scheme is based on learning from the past experience, we expect that it will be able to satisfy QoS requirements, in a variety of realistic scenarios, which can not be accounted for in analytical models.
I. I NTRODUCTION
In this paper we deal with the problem of Radio Admission Control (RAC) for various service types in Long Term
Evolution (LTE) system. While in previous cellular systems, such as 3G, voice traffic is transferred through circuit-switched networks, in LTE is transferred through packet-switched networks along with best-effort data services (such as FTP and
HTTP traffic). Unlike best-effort services, Voice over IP (VoIP) traffic has very tight Quality of Service requirements, such as bit rate and delay. In order to support diverse services which have specific QoS requirements, dedicated bearers are set up within Evolved Packet System (EPS), where each bearer is an
IP packet flow associated
A Cognitive Scheme for Radio Admission Control in LTE systems
Biljana Bojovic, Nicola Baldo and Paolo Dini
Centre Tecnol`ogic de Telecomunicacions de Catalunya (CTTC), IP Technologies Area
Av. Carl Friedrich Gauss 7, Castelldefels (Barcelona), Spain
{bbojovic, nbaldo, pdini}@cttc.es
Abstract— In order to provide QoS requirements in high speed future communication networks, such as LTE, the operator has to provide a Radio Admission Control algorithm which will guarantee the QoS of different service types (e.g. voice, data, video, ftp) while maximizing radio resource utilization. In this paper we propose a Cognitive Radio Admission Control Scheme based on a Multilayer Feed-forward Neural Network. According to our scheme, the eNodeB performs Radio Admission Control using a cognitive engine that learns from the past experience how the admission of a new session would affect the QoS of all sessions in the future. We implemented our Radio Admission
Control Scheme using a LTE-EPC simulator. Since our scheme is based on learning from the past experience, we expect that it will be able to satisfy QoS requirements, in a variety of realistic scenarios, which can not be accounted for in analytical models.
I. I NTRODUCTION
In this paper we deal with the problem of Radio Admission Control (RAC) for various service types in Long Term
Evolution (LTE) system. While in previous cellular systems, such as 3G, voice traffic is transferred through circuit-switched networks, in LTE is transferred through packet-switched networks along with best-effort data services (such as FTP and
HTTP traffic). Unlike best-effort services, Voice over IP (VoIP) traffic has very tight Quality of Service requirements, such as bit rate and delay. In order to support diverse services which have specific QoS requirements, dedicated bearers are set up within Evolved Packet System (EPS), where each bearer is an
IP packet flow associated