Introduction

In recent years, wireless networks, especially ad hoc networks that consist of a collection of radio transceivers without requiring centralized administration or a prearranged fixed network infrastructure, have been studied intensively. Considering the application scenarios in which the users are “selfish” and act noncooperatively to maximize their own interests, the performances of such networks will deteriorate dramatically because of the inefficient competition for the wireless resources among selfish users. The greediness of selfish users and the distributed network structure challenge the feasibility of the conventional approaches and require novel techniques for distributed and efficient networking. Thus ensuring cooperation among selfish users becomes an important issue for designing wireless networks.

To ensure the cooperation and study the behaviors of selfish users, game theory is a successful economy tool, which studies the mathematical models of conflict and cooperation between intelligent and rational decision makers. In the literature, different types of game approaches have been introduced to several areas of wireless communications. One of the most important is the pricing anarchy, in which a price is taxed for the resource usage so that cooperative behaviors can be enforced. Noncooperation game theory was studied in [268] for power-control problems, in which the pricing technique was used to achieve Pareto optimality. In [369], resource allocation was studied for a forward link two-cell CDMA voice network with multiple service classes. Noncooperative game theory has also been studied for self-organizing mobile ad hoc wireless networks (MANET). In [38, 368], reputation-based game approaches were proposed to encourage packet forwarding among users.

Because of fading channels, user mobility, energy/power resources, and many other factors, cross-layer design and multiuser optimization are the keys to ensuring overall system performance of wireless networks. And resource allocation is one of the most important issues for implementing future wireless networks.

In the past decade, we have witnessed significant progress in the advance of resource allocation over wireless networks. It is not only an important research topic, but is also gradually becoming an integral teaching material for graduate-level networking courses.

Yet there are few books available to date that can serve such a purpose. Why? Because the field of resource allocation is such a versatile area that covers a broad range of issues, it is not easy to develop a comprehensive book to cover them all. For instance, resource allocation across various networking layers encounters different design constraints and parameters; different networking scenarios have different performance goals and service objectives; and different formulations of resource allocations need to employ different optimization tools.

To respond to the need of such a book for graduate students, researchers, and engineers, we try to tackle the difficulties by bringing together our research in resource allocation over the past decade and the basic material of resource allocation and optimization techniques to form the foundation of this book. Its intent is to serve either as a textbook for advanced graduate-level courses on networking or as a reference book for self-study by researchers and engineers.

This book covers three main parts. In Part I, the basic principles of resource allocation is discussed.

With the advancement of multimedia compression technology and wide deployment of wireless networks, there is an increasing demand especially for wireless multimedia communication services. The system design has many challenges, such as fading channels, limited radio resources of wireless networks, heterogeneity of multimedia content complexity, delay and decoding dependency constraints of multimedia, mixed-integer optimization, and trade-offs among multiuser service objectives. To overcome these challenges, dynamic resource allocation is a general strategy used to improve the overall system performance and ensure individual QoS. Specifically in this chapter, we consider two aspects of design issues: cross-layer optimization and multiuser diversity. We study how to optimally transmit multiuser multimedia streams, encoded by current and future multimedia codecs, over resource-limited wireless networks such as 3G cellular systems, WLANs, 4G cellular systems, and future WLAN/WMANs.

Introduction

Over the past few decades, wireless communications and networking have experienced an unprecedented growth. With the advancement in multimedia coding technologies, transmitting real-time encoded multimedia programs over wireless networks has become a promising service for such applications as video-on-demand and interactive video telephony. In most scenarios, multiple multimedia programs are transmitted to multiple users simultaneously by sharing resource-limited wireless networks.

The challenges for transmitting multiple compressed multimedia payloads (such as videos) over wireless networks in real time lie in several factors. First, wireless channels are impaired by detrimental effects such as fading and CCIs. Second, there are limited radio resources, such as bandwidth and power, in the wireless networks.

Multi-hop relaying routing protocols have been investigated for CDMA air interfaces in conventional cellular scenarios, as described in Chapter 6 and in (Harold and Nix, 2000a) (Harold and Nix, 2000b). This chapter compares the performance of ODMA with direct transmission for cases where links maybe required directly to other nodes, as well as to a controlling (backbone) node, and presents two new routing algorithms.

For an interference-limited system, it is shown that the topology cannot be supported by a conventional (single-hop) system, but that a relayed system is able to provide service. As an enhancement to path loss routing, a new admission control and routing algorithm based on receiver interference is presented which is shown to further enhance performance.

A second new routing algorithm, which considers the interaction between all receivers in the system by means of a ‘congestion’ measure is presented. This approach allows for routing that is optimised for the entire system, not just a particular route under arbitrary starting conditions. This is possible under both central and local parameter gathering scenarios. Through formulating this measure into the power control equations it is possible to determine system feasibility, although this is a conservative criterion due to approximations in the formulation. This congestion-based routing is shown to outperform non-relaying and any previous routing technique in available capacity for the new network topologies, and has the lowest transmitted power requirement of all investigated methods.

Introduction

A special property of the TDD mode is that not only the mobile stations (MSs) can interfere with the base stations (BSs), but also the MSs may interfere with each other. The same holds for BSs, which can interfere with MSs as well as other BSs. This property of a TDD interface creates complex interference situations. The resulting interference has a significant impact on capacity. In this chapter, other-cell interference including adjacent-channel interference (ACI) and co-channel interference (CCI) are examined. In this context, multi-operator scenarios are considered since in UMTS (Universal Mobile Telecommunications System), for instance, any one operator may only obtain a single TDD carrier.

In order to assess the impact of interference on capacity (the number of users that can be served simultaneously), in section 3.2 a new equation is derived to calculate the system capacity relative to the non-interfered state (a single, isolated cell). In addition, the pole capacity (the theoretical upper bound of capacity) is calculated for the case of ideal power control and non-ideal power control.

In section 3.3 ACI is studied in single-cell and multiple-cell environments. The probability density function (pdf) of interference in the single-cell environment is derived analytically in order to validate the simulation platform. In the multiple-cell environment the effects of handover and power control on interference and capacity, respectively, are examined.

In section 3.4 CCI is analysed with a special focus on distribution of interference with respect to the basic interference sources: MSs and BSs.

Introduction

Overview of topic

The last two decades have been witness to the rapid growth and widespread success of wireless communications. This is largely due to breakthroughs in communication theory and progress in the design of low-cost power-efficient mobile devices. Third generation (3G) and other future systems (Nakajima and Yamao, 2001) must provide transmission at a significantly higher data rates as well as a lower tariff per transmitted bit than current second generation (2G) systems. It should be noted that wireless technology is becoming more familiar and common for society. They will also need to address the issue of how many mobiles can be simultaneously served with the infrastructure that the provider has in place. It is clear that we will witness also newer modes and improved services of communication due to progress in radio technology.

The requirements on data rate, link reliability, spectral efficiency and mobility in a fading environment cannot be met with conventional single antenna systems. Therefore, antenna arrays may be used at least on the base station (BS) side to improve performance. Indeed, the new Chinese time division duplex (TDD) system time-division synchronous CDMA (Yang et al., 2002) has been specifically designed for use with smart antenna arrays. Antenna arrays can be used to provide an increased antenna gain and/or an increased diversity gain for each user in the system. At the same time, less interference is received from the other directions (on the uplink) or transmitted to other mobiles (on the downlink).

Introduction

Over the last 20 years personal computers have developed as a mass consumer product which, coupled with the growth in widespread broadband access, offers users a multiplicity of services. The Internet is an enormous source of information (both valid and invalid) that has evolved into a virtual store, a library, a chat-room and a playground, and has enabled novel means of interacting. The simple email service is an example of how material streams (transport of letters, documents, etc.) have been replaced by information streams. These issues are leading to fundamental changes in how we do things.

A further significant step in the ‘digital revolution’ has been the demand for mobility, the so called anyone, anywhere, anytime mentality. Over the world digital mobile telephony has been a huge success. In particular, in Europe, the foundation was laid when 13 countries agreed to adopt a single digital standard for cellular mobile communications (The Memorandum of Understanding (MoU) signed on 7 September 1987 in Copenhagen). The development of the Global System for Mobile communications (GSM) followed, and the historic milestone of 1 billion mobile subscribers was reached in 2000 (Eylert, 2000). It is predicted that by the year 2010 there will be more than 1.7 billion terrestrial mobile subscribers worldwide.

Motivation and problems

High peak data rate transmission, network self-organisation and universal frequency reuse are considered important features for future cellular, ad hoc, multi-hop and hybrid wireless networks (Prehofer and Bettstetter, 2005). OFDMA is viewed as a promising modulation/multiple-access technique for providing very high data-rates and flexible resource allocation while at the same time enabling low complexity receivers (Stimming et al., 2005). Time-division duplexing (TDD) supports traffic asymmetry very well which is inherent to packet data services. Moreover, TDD offers channel reciprocity which is exploited in this research in a novel fashion for medium access and subchannel allocation. The problems that arise from TDD are the requirement for time synchronisation and additional interference scenarios. This is particularly important as OFDMA performs poorly under conditions of universal frequency reuse because of the high CCI. Figure. 12.1 illustrates the CCI problem in a cellular network using TDD with frequency reuse of one. The figure shows two adjacent BSs (base stations), namely BS1 and BS2 with a MS (mobile station) associated with each BS, namely MS1 and MS2. MS1 is transmitting data to BS1. Consequently, MS1 causes interference to MS2, since MS2 is in receiving mode. Similarly, BS2 causes interference to BS1. Due to the potentially small spatial separation between transmitter and ‘victim’ receiver and the low path loss between BSs due to line-of-sight conditions, in a full frequency reuse network, CCI poses a major challenge on the MAC protocol design and channel assignment procedure.

Introduction

With the advent of third generation mobile communication systems TDD technology was applied for the first time to cellular systems. These systems primarily make use of CDMA (code-division multiple access) for managing multiuser access. Therefore, the scope of this chapter is to study the combination of CDMA and TDD technologies applied to cellular systems.

The capacity of cellular systems that use CDMA in combination with frequency-division duplex (FDD) have been intensively studied (Chebaro and Godlewski, 1992; Wu et al., 1997; Corazza et al., 1998; Veeravalli and Sendonaris, 1999; Takeo et al., 2000) following the pioneering papers by (Gilhousen et al., 1991) and (Viterbi and Viterbi, 1994). Gilhousen and Viterbi applied spread-spectrum techniques, which, until then, had primarily been used in military applications, to commercial wireless communication systems — the foundation for which was laid by Shannon's communication theory (Shannon, 1948). Due to its flexibility in supporting various transmission rates, CDMA had been widely adopted for third-generation mobile communication systems. During the European FRAMES (future radio wideband multiple access system) project in the mid nineties, different air interface proposals were investigated including a hybrid TD-CDMA/FDD system (delta concept), results of which are reported in (Ojanperä et al., 1997), (Pehkonen et al., 1997), and (Klein et al., 1997).

The time-division duplex (TDD) technique has been applied successfully to frequency-division multiple access (FDMA) systems and time division-multiple access (TDMA) systems (Esmailzadeh et al., 1997; Horikawa et al., 1997).

Introduction

This chapter aims to exploit the key finding of Chapter 3 and apply it to the TDD air interface of UMTS (UTRA-TDD). The significant finding of the previous chapter is that ideal synchronisation is not necessarily a prerequisite to obtaining the maximum capacity in a TD-CDMA/TDD network. This has led the author to develop a novel technique which is called the time slot (TS)-opposing method. In this chapter this method is used to develop a centralised dynamic channel assignment (DCA) algorithm.

The approach in this chapter is as follows: first, in section 4.2 a simple centralised DCA algorithm used in a single cell is studied. This investigation aims to find an upper bound of the network performance when combining the TS-opposing technique with a DCA algorithm. Second, in section 4.3 the TS-opposing algorithm is investigated in a cellular TD-CDMA/TDD network. For this approach it is assumed that a group of BSs (following the so called bunch concept (Mihailescu et al., 1999)) is connected to a radio network controller (RNC).

TS-opposing technique applied to a single cell

In this investigation an idealised deployment scenario is assumed to investigate the new TS-opposing mechanism. This means that a TS-opposing algorithm is employed with the aim of improving the capacity only with respect to a single cell. The capacity obtained thereby is then compared with the capacity of an equivalent FDD interface.

A cluster of seven hexagonal cells is assumed with the cell of interest (COI) in the centre.

Until now, the focus of the book has been on interference analysis and management for either a cellular TDD-CDMA system or an ODMA enhanced TDD-CDMA system. Interference management can be viewed as a form of resource management. In this chapter the issue of link adaptation is addressed with a focus on what metrics are appropriate to enable radio resource management in a cross-layer manner. In conventional systems, the decision on how to choose the ideal physical mode (PHY-mode) is primarily based on the knowledge of the interference encountered. This information is reported to higher-layer entities that deal with the radio resource management. Subsequently, the radio resource management entity makes a decision as to which physical resource will be used (e.g. by employing a DCA algorithms as described in Chapter 5 and 8) and this information is reported back to the physical layer. It is apparent that this process can take quite a long time. Meanwhile the channel conditions may have changed significantly. Hence, the previously chosen radio resource may no longer be the ideal choice. Therefore, new methods (e.g. resource metric estimation) are discussed here that (a) base the decision as to which radio resource to use, not only on interference, but also, for example, on the statistics of the channel state information, and (b) make the decision as to which radio resource to use already at the physical layer. For (a) it is shown that the TDD mode is ideally suited due to the reciprocity of the channel.

DCA techniques

Congestion-based routing, as developed in the previous chapter, is shown to require the lowest transmitted power, and in most cases achieves the highest capacity of all the routing algorithms examined in Chapter 5. All of these routing algorithms have allocated TDD time slots on a first-come-first-served basis and according to the rules outlined in section 6.4.1. This allocation only serves to ensure that the limitations of the TDD hardware are considered. It makes no attempt to optimise time slot allocation.

The allocation of time slots with regard to system performance has been shown to be an effective technique to mitigate interference (Haas, 2000). Integrating slot allocation, or DCA, into the routing algorithm would appear to be the most effective approach due to the interactive nature of interference. This approach will also need to conform to the extra limitations imposed by relaying. In addition to the rules in section 6.4.1, it is obvious that the slot allocation must be in the same order as the relays. A combined DCA will allow minimisation of the desired measure, in this case congestion, simultaneously in routing and slot allocation. This chapter develops a combined routing and resource allocation algorithm for TDD-CDMA relaying. It starts by reviewing one such algorithm applicable to TDMA and FDMA. A novel method of time slot allocation according to relaying requirements is then developed. Two measures of assessing congestion are presented based on matrix norms.

In this and the following two chapters the focus moves away from networks which are controlled centrally by a base station to a hybrid cellular network which permits cellular operation as well as peer-to-peer operation. Essentially we consider multi-hop wireless networks based on opportunity-driven multiple access (ODMA) which will be shown to reduce the overall transmission power in a system, to be resilient to shadowing and to potentially increase the coverage compared with single-hop transmission. However, for simple receivers and low user density, the actual capacity of UTRA-TDD may be marginally reduced from the maximum non-relaying capacity. This chapter begins the study of ODMA based systems by analysing the implications of relaying in a cellular scenario compared to a conventional nonrelaying system. Initially the interference is analysed by investigating the effect of reduced transmitted power resulting from reduced path loss for a link. The effect of shadowing is considered and it is shown that a relaying system is able to benefit from increased zero mean lognormal shadowing by utilising the diversity of paths available. A correlated shadowing model is developed from a previous model considering both distance and angle of arrival (Klingenbrunn and Mogensen, 1999) to include the shadowing correlation between all transceivers, as they may all be available to receive in a relaying environment. It is shown that while this affects the interference pattern the perturbation is not significant.

Introduction

In this chapter a distributed dynamic channel assignment (DCA) algorithm applicable for the TDD mode of the UMTS terrestrial radio access (UTRA) is presented. It is closely related to the DCA used in the DECT (digitally enhanced cordless telecommunications) system (Punt et al., 1998). Once again, the discovery made in Chapter 3 is exploited; that is, that for certain scenarios opposed synchronisation of TSs between neighbouring cells is advantageous. The new distributed DCA algorithm is supported by the results of the investigation in section 4.2. In this section it was demonstrated that synchronous transmission and reception between neighbouring cells may not yield the greatest capacity that is attainable in a single cell. It was found, however, that when the centralised DCA algorithm developed in section 4.3 was applied to multiple cells it was not feasible to fully exploit the potential gains revealed by the capacity analysis of a single cell.

In this chapter it is demonstrated that by applying the novel distributed DCA algorithm, which utilises the TS-opposing idea, greater capacity can result than would be obtained by synchronous transmissions. Most importantly, this is shown to be valid for a TDMA-CDMA/TDD (TD-CDMA/TDD) network, which accounts for full spatial coverage. Channel asymmetry is assumed to be arranged by code pooling rather than TS pooling (3GPP, TSG, RAN, 2000c).