1994 IEEE TRANSACTIONS ON WIRELESS COMMUNICATIONS, VOL. 14, NO. 4, APRIL 2015 Performance Analysis of a Connection Admission Scheme for Future Networks Tom Mmbasu Walingo, Member, IEEE, and Fambirai Takawira, Member, IEEE Abstract—Future networks are to deliver any-traffic, anytime, anywhere with full quality of service (QoS) guarantees. They will evolve from typical heterogeneous networks (HetNets) into dense, organic, and irregular heterogeneous networks called DenseNets. They will be complex and face additional challenges of heterogene- ity in many design dimensions like different radio access technolo- gies (RAT’s) shrinking in structure. Radio Resource Management (RRM) is one of the key challenges in providing for QoS for these networks. Connection Admission Control (CAC) scheme and intel- ligent scheduling techniques are employed on the links for RRM. In this paper a CAC scheme is developed that features multiple traffic classes, multiple admission parameters at both packet and connection level. The CAC scheme uses both signal to interference ratio (SIR) and delay as admission parameters since the single parameter based CAC algorithm is not adequate for the emerging traffic classes. The performance analysis of the model features Batch Markovian Arrival Process (BMAP) traffic, a better rep- resentative of the future traffic characteristics than the traditional Poisson traffic. A simple approximate Markovian analytical model is developed and used to analyze the complex network. The devel- oped model with more admission parameters outperforms those with less admission control parameters for future networks traffic. Index Terms—Batch Markovian Arrival Process (BMAP) traf- fic, connection admission control (CAC), Code Division Multiple Access (CDMA), dense networks (DenseNets), heterogeneous net- works (HetNets), multimedia traffic. I. I NTRODUCTION F UTURE networks are to deliver any traffic anytime any- where with full quality of service (QoS) guarantees. They will evolve from typical heterogeneous networks (HetNets) into dense, organic, and irregular heterogeneous networks called DenseNets. These multiservice wireless/mobile broad- band networks will be heterogeneous with the following forms of diversity; network diversity, terminal diversity and traffic/ applications diversity. The networks will encompass diverse ir- regular cell sizes in a multi-tier configuration and access points with different characteristics and technologies in their operating Manuscript received August 1, 2013; revised December 16, 2013, May 19, 2014, and November 16, 2014; accepted November 19, 2014. Date of publication December 8, 2014; date of current version April 7, 2015. This work was supported in part by Telkom and Alcatel-Lucent as part of the Center of Excellence Programme at the Center of Radio Access and Rural Technologies, University of KwaZulu-Natal. The associate editor coordinating the review of this paper and approving it for publication was A. Wyglinski. T. M. Walingo is with the Centre of Radio Access and Rural Technolo- gies, University of KwaZulu-Natal, Durban X54001, South Africa (e-mail: Walingo@ukzn.ac.za). F. Takawira is with the School of Electrical and information Engineering, University of the Witwatersrand, Johannesburg 2050, South Africa (e-mail: Fambirai.takawira@wits.ac.za). Color versions of one or more of the figures in this paper are available online at http://ieeexplore.ieee.org. Digital Object Identifier 10.1109/TWC.2014.2378777 environments [1], [2]. The networks are faced with the chal- lenges associated with increasing the capacity, small closely packed cell sizes, accommodating diverse heterogeneous ac- cess networks, mobility management and dealing with diverse network protocols, one of them being Code Division Multiple Access (CDMA) and its variants with its own inherent prob- lems [3]. Apart from high intelligence, future terminals will feature multiple interface capabilities to share traffic load on different networks simultaneously (multi-homed terminals) or one network at a time (multi-mode terminals). The network’s Radio Resource Management (RRM) decides the Radio Access Technology (RAT) a terminal connects to. The network traffic and applications are diverse in their quality of service (QoS) requirements, multi-class and are ever growing, these include [4]; operator-consumer applications e.g., mobile television, peer-to-peer applications (e.g., instant messaging, voice-over- ip, video conferencing), machine-to-machine applications (e.g., data telemetry and automotive applications), mobile web ser- vices (e.g., music and video streaming), and social networking applications, all these with their own challenges. The reduc- tion in cell sizes, the increase in number of diverse terminals and tremendous traffic volumes of the networks will result in very closely packed irregular heterogeneous dense networks (DenseNets) [5]. Radio Resource Management (RRM) is one of the key challenges in providing for QoS for these networks. Connection Admission Control (CAC) schemes and intelligent scheduling techniques are employed on the links for RRM. The RRM/CAC decides whether to accept an incoming connection and selects the best available RAT for the connection request. The most popular CAC/RAT selection schemes include the load balanc- ing schemes where a connection is assigned to a RAT until the maximum RAT’s determined load is reached [6]–[8] and the utility based schemes where utility functions numerically quantify the factors taken into account in the RAT selection de- cision as determined by user satisfaction, these could be user’s QoS parameters [8]–[10]. In most systems e.g., CDMA with performance degradation as more users enter the system, signal to interference ratio (SIR) regulates the network load. In fact, most parameters used in the CAC schemes e.g., cost, path loss, throughput, bit error rate, waiting time etc., are directly or in- directly influenced by the networks signal-to-interference-plus noise ratio (SINR) or SIR. In [7], “Myth 1: The received signal- to-interference-plus noise ratio (SINR) is the first-order predic- tor of the user experience or at least of the link reliability”, indicates the need to consider network load with SINR. How- ever, the load in the system self regulates due to performance degradation. Therefore, SIR is still the best general indicator 1536-1276 © 2014 IEEE. Personal use is permitted, but republication/redistribution requires IEEE permission. See http://www.ieee.org/publications_standards/publications/rights/index.html for more information.