Call admission control for wireless personal communications
Introduction
The next generation wireless personal communication systems (PCS) are expected to revolutionize the way in which people communicate [4], [7], [9], [14], [22], [29], [45], [54], [73], [74]. The revolution is brought about by allowing people to communicate with anyone, in any place, at any time, in a multimedia environment and with a pre-specified level of quality-of-service (QoS). Because of scarce radio spectrum and limited coverage of base stations, the wireless multimedia networks are connected to wireline networks to maximize the coverage area of the PCS [1], [79]. That is, the PCS will be made up of interconnected regional wireless and wireline systems, as shown in Fig. 1, where WNRCS stands for Wireline Regional Communication System and WSRCS stands for Wireless Regional Communication System. A region can be a city, a province or state, or a country. The coverage areas of some of the regional systems can be overlapped, and far apart ones can be interconnected by high capacity and reliable dedicated links. These dedicated connections can be through satellites or under-ocean cables, or simply overland. The aggregate of all interconnected wireline regional systems can be treated as a giant backbone. Because wireline systems are more stable, much easier to upgrade and maintain, and have more capacity, the backbone will be a very important part of the PCS, acting as (i) a reliable passageway for separated wireless systems, and (ii) a connection point to large information database in the wireline domain. The backbone can be either ATM or IP based.1
QoS provisioning is a major feature of the future PCS. In most systems, call admission control (CAC) is the first control function imposed on a user for QoS provisioning. When a user requests a new connection, what the CAC usually does is to calculate the amount of resources required by (i) the users already in the system, and (ii) the pending user. If the sum of the two is not more than the total capacity, then the user's request will be acknowledged; otherwise the request will be rejected. This is equivalent to first reserving resources for the admitted users and then checking if the remaining resources are sufficient to support the new connection. How to make the two calculations depends on the particular system under consideration. CAC is important because its result is irreversible. A CAC mechanism is usually defined as the detailed work involved in the CAC function. This includes the decision process, signaling, routing table establishment, etc. The decision process of CAC can often be formulated in a high level representation called the CAC policy.2 Whenever a user requests a new connection, the CAC policy takes the call request as input, and based on the current traffic conditions of the system, decides whether or not to accept the user, as illustrated in Fig. 2.
Established in the PCS environment, the entire end-to-end connection of each call, with the possibility of spanning both the wireline and the wireless domains, may consist of wireline hop(s) and wireless hop(s). During the call admission phase, the decision process of CAC is repeated over all the hops to ensure that sufficient resources are available to support the entire new connection. A significant amount of work on CAC has been done for the wireline domain alone [13], [16], [27], [28], [34], [36], [39], [40], [47], [64], [78], [80], for the wireless domain alone [17], [53], [67], and for both [2], [3], [19], [21], [49], [65], [66], [85]. The works on CAC for connection oriented wireline networks (such as ATM networks) are more matured and there are usually less challenges involved. However, CAC in wireless, IP-based wireline, and interconnected wireless/wireline networks poses significant technical challenges due to user mobility, limited radio spectrum, dynamic nature of multimedia traffic, hostile wireless propagation environments, IP connectionless nature, etc. As a result, a robust CAC policy that can facilitate the provision of QoS in the PCS is hard to find. Most recent efforts on CAC are dedicated to the wireless portion of the end-to-end connection. This paper is aimed at providing a survey of the current existing proposals for CAC in the PCS, with emphasis placed on CAC in the wireless domain and in the combined wireless/wireline IP domain. In particular, the paper will provide a review of the previous works on CAC reported in [2], [3], [17], [19], [20], [21], [49], [53], [65], [66], [67], [85]. The survey will help to identify what problems have been and have not been studied. Various efforts aimed at a specific problem will be discussed and comments will be made in order to deliver a comprehensive insight to the previous studies. This is to provide an overview on CAC to general audience and, at the same time, to assist perspective researchers to focus their efforts on untackled problems.
User mobility is the most critical aspect that must be addressed in any literature related to wireless communications. Virtual connection tree (VCT) and cell cluster [1], [3] are common strategies used to handle the mobility in PCS. The VCT and the cell cluster structures in a wireless network can be classified into two categories, namely the static VCT static cluster scheme and the dynamic VCT dynamic cluster scheme. The CAC policies for systems using each of the schemes will be discussed. The remainder of this paper is organized as follows. Section 2 presents the major challenges in CAC for personal wireless communications. Section 3 describes the system infrastructures proposed to handle user mobility. It gives some important assumptions made in the literature and the definitions of the VCT and the cell cluster. The works on CAC in a system using the static scheme are reviewed in Section 4, while the works on CAC in a system using the dynamic scheme are reviewed in Section 5. Section 6 briefly reviews some preliminary works on CAC in wireless/IP interworking. Finally, Section 7 summarizes this survey and identifies the issues that should be investigated in the future.
Section snippets
Major challenges in CAC
There are many challenges in the wireless and the combined wireless/wireline environments. The limited radio spectrum, user mobility and fluctuation of usable bandwidth due to time-varying channel conditions are typical stumbling blocks found in such environments. Each of these problems is separately addressed in [26], [37], [44], [56], [77], [33], [60], [61] and [5], [68], respectively. In addition, the complex interaction between the wireless and wireline networks [19], [68] and the
System infrastructure
CAC is a function closely related to the infrastructure of the system. Because of this, an interconnection structure for the wireless and wireline domains is required before the CAC functions can be implemented. Also, because of the convenience brought about by tetherless connections, the number of active mobile users in the future PCS is expected to be large. As a result, there will be a lot of information flowing between the wireless and the wireline domains. A carefully designed
The static VCT static cluster structure
The arrangements of the VCT and the cell cluster in [2], [3], [17], [21], [65], [66] are similar to each other. For illustration, consider a single wireless regional system and the backbone. The coverage area of the wireless regional system is first divided into fixed subregions. Each of these subregions can be looked at as a cell cluster. For each subregion, dedicated wireline links are then used to connect together all the base stations within, via a central wireline switch. The bunch of
The dynamic VCT dynamic cluster structure
The work on CAC in [49], [53], [67], [85] will be examined together in this section, on the basis that the arrangements of the VCT and the cell cluster in these papers are similar to each other. In particular, the arrangement scheme used in these papers is different from the static scheme described in Section 4 in three aspects. First of all, in the dynamic scheme a VCT and a cell cluster is set up for every single mobile user admitted to the wireless network [53], [85]. Secondly, the sizes of
Wireless/IP interworking
The pervasiveness of the Internet and the flexibility of the wireless communication network to support user roaming make ‘interworking’ of these two information transport platforms imperative for the support of multimedia services between remotely located mobile users. By design, the Internet only offers best effort service for fixed users, while the wireless environment supports user mobility and is prone to noise and interference. The interworking of these two communications subnets for
Conclusions
Interconnection of the wireless and wireline domains using VCTs and cell clusters is geared to the solution of wireline handoff. This particular advantage comes from the fact that the major access point to the wireline backbone is fixed, making an end-to-end connection easier to maintain. According to the resource arrangement scheme used, the interconnection structure can be divided into two categories: the static VCT static cluster scheme and the dynamic VCT dynamic cluster scheme.
The works on
Acknowledgements
This work was partially supported by Communications and Information Technology Ontario (CITO) and by Natural Sciences and Engineering Research Council of Canada (NSERC).
References (85)
- et al.
Admission control and routing in ATM networks
Comput. Networks ISDN Syst.
(1990) - et al.
Optimal call admission policy for wireless multimedia networks
Comput. Commun.
(2002) Wireless ATM: a perspective on issues and prospects
IEEE Pers. Commun.
(1996)- et al.
Control and quality-of-service provisioning in high-speed microcellular networks
IEEE Pers. Commun.
(1994) - et al.
An architecture and methodology for mobile-executed handoff in cellular ATM networks
IEEE J. Select. Areas Commun.
(1994) - et al.
SWAN: a mobile multimedia wireless network
IEEE Pers. Commun.
(1996) - et al.
Adaptive mobile multimedia networks
IEEE Pers. Commun.
(1996) MPLS: the magic behind the myths
IEEE Commun. Mag.
(2000)- et al.
Trends in standardization on multimedia communications
IEEE Commun. Mag.
(1997) - et al.
Optimal scheduling of handoffs in cellular networks
IEEE/ACM Trans. Networking
(1996)