The main drawback of the multi-node decode-and-forward (DF) protocol and the multinode amplify-and-forward (AF) protocol, presented in Chapter 5, is the loss in the data rate as the number of relay nodes increases. The use of orthogonal sub channels for the relay node transmissions, either through TDMA or FDMA, results in a high loss of the system spectral efficiency. This leads to the use of what is known as distributed space–time coding (DSTC) and distributed space–frequency coding (DSFC), where relay nodes are allowed to simultaneously transmit over the same channel by emulating a space–time or a space–frequency code. The term distributed comes from the fact that the virtual multi-antenna transmitter is distributed between randomly located relay nodes. Employing DSTCs or DSFCs reduces the data rate loss due to relay nodes transmissions without sacrificing the system diversity order, as will be seen in this chapter.

Distributed space–time coding (DSTC)

In this section, the design of distributed space–time codes for wireless relay networks is considered. The two-hop relay network model depicted in Figure 7.1, where the system lacks a direct link from the source to destination node, is considered. Distributed space–time (space–frequency) coding can be achieved through node cooperation to emulate multiple antennas transmitter. First, the decode-and-forward protocol, in which each relay node decodes the symbols received from the source node before retransmission, is considered.

Wireless communications have seen a remarkably fast technological evolution. Although separated by only a few years, each new generation of wireless devices has brought significant improvements in terms of link communication speed, device size, battery life, applications, etc. In recent years the technological evolution has reached a point where researchers have begun to develop wireless network architectures that depart from the traditional idea of communicating on an individual point-to-point basis with a central controlling base station. Such is the case with ad-hoc and wireless sensor networks, where the traditional hierarchy of a network has been relaxed to allow any node to help forward information from other nodes, thus establishing communication paths that involve multiple wireless hops. One of the most appealing ideas within these new research paths is the implicit recognition that, contrary to being a point-to-point link, the wireless channel is broadcast by nature. This implies that any wireless transmission from an end-user, rather than being considered as interference, can be received and processed at other nodes for a performance gain. This recognition facilitates the development of new concepts on distributed communications and networking via cooperation.

The technological progress seen with wireless communications follows that of many underlying technologies such as integrated circuits, energy storage, antennas, etc. Digital signal processing is one of these underlying technologies contributing to the progress of wireless communications. Perhaps one of the most important contributions to the progress in recent years has been the advent of MIMO (multiple-input multiple-output) technologies.

In this chapter, we present a cooperative protocol based on the relay-selection technique using the availability of the partial channel state information (CSI) at the source and the relays. The main objective of this scheme is to achieve higher bandwidth efficiency while guaranteeing the same diversity order as in a conventional cooperative scheme. Two cooperation scenarios are addressed: a single-relay scenario and a multi-relay scenario. In the single-relay scenario, the focus is to answer the question: “When to cooperate?” The rationale behind this protocol is that there is no need for the relay to forward the source's information if the direct link, between the source and destination, is of high quality. It turns out that an appropriate metric to represent the relay's ability to help is a modified version of the harmonic mean function of its source–relay and relay–destination instantaneous channel gains. The source decides when to cooperate by taking the ratio between the source–destination channel gain and the relay's metric and comparing it with a threshold, which is referred to as the cooperation threshold. If this ratio is greater than or equal to the cooperation threshold, then the source sends its information to the destination directly without the need for the relay. Otherwise, the source employs the relay in forwarding its information to the destination as in the conventional cooperative scheme.

The previous chapter studied the effects and use of cooperation on the multiple access channel. In this chapter we look further into using the properties of the source traffic to improve the efficiency of cooperative multiple access. Because the presentation in this chapter is highly dependent on the characteristics of the source, we will focus on the communication of packet speech. Nevertheless, the main underlying ideas can be extended to other types of sources.

Speech communication has a distinctive characteristic that differentiates it from data communication, which was the main focus of the previous chapter. Speech sources are characterized by periods of silence in between talk spurts. The speech talk–silence patterns could be exploited in statistical multiplexing-like schemes where silent users release their reserved channel resources, which can then be utilized to admit more users to the network. This comes at the cost of requiring a more sophisticated multiple access protocol. One well-known protocol that uses this approach for performance improvement is the packet reservation multiple access (PRMA) protocol [49], which can be viewed as a combination of TDMA and slotted ALOHA protocols. In PRMA, terminals in talk spurts contend for the channel in empty time slots. If a user contends succesfully, then the slot used for contention is reserved for the user. Users with reservations transmit their voice packets in their reserved slots.

In the previous chapters, the gains of cooperative diversity were established under the ideal model of negligible listening and computing power. In sensor networks, and depending on the type of motes used, the power consumed in receiving and processing may constitute a significant portion of the total consumed power. Cooperative diversity can provide gains in terms of savings in the required transmit power in order to achieve a certain performance requirement because of the spatial diversity it adds to the system. However, if one takes into account the extra processing and receiving power consumption at the relay and destination nodes required for cooperation, then there is obviously a tradeoff between the gains in the transmit power and the losses due to the receive and processing powers when applying cooperation. Hence, there is a tradeoff between the gains promised by cooperation, and this extra overhead in terms of the energy efficiency of the system should be taken into consideration in the design of the network.

In this chapter the gains of cooperation under this extra overhead are studied. Moreover, some practical system parameters, such as the power amplifier loss, the quality of service (QoS) required, the relay location, and the optimal number of relays, are considered. Two communications architectures are considered, direct transmission and cooperative transmission. The performance metric for comparison between the two architectures is the energy efficiency of the communication scheme.

In broadband communications, OFDM is an effective means to capture multipath energy, mitigate the intersymbol interference, and offer high spectral efficiency. OFDM is used in many communications systems, e.g., wireless local area networks (WLANs) and wireless personal area networks (WPANs). Recently, OFDM together with time–frequency interleaving across subbands, the so-called multiband OFDM [9], has been adopted in the ultra-wideband (UWB) standard for wireless personal area networks (WPANs).

To improve the performance of OFDM systems, the fundamental concept of cooperative diversity can be applied. Nevertheless, special modulations/cooperation strategies are needed to efficiently exploit the available multiple carriers.

In this chapter, we study an OFDM cooperative protocol that improves spectral efficiency over those based on fixed relaying protocols while achieving the same performance of full diversity. By exploiting limited feedback from the destination node, the described protocol allows each relay to help forward information of multiple sources in one OFDM symbol. We also describe a practical relay assignment scheme for implementing this cooperative protocol in OFDM networks.

System model

In this section, we describe the system model of a wireless network, in which we take into consideration the random users’ spatial distribution. The channel model, the signal model, and the performance measure in term of outage probability are discussed.

We consider an OFDM wireless network such as a WLAN or a WPAN with a circular cell of radius ρ. The cell contains one central node and multiple users, each communicating with the central node.

For practical implementation of cooperative communications in wireless networks, we need to develop protocols by which nodes are assigned to cooperate with each other. In most of the previous chapters on cooperation, the cooperating relays are assumed to exist and are already paired with the source nodes in the network. A deterministic network topology, i.e., deterministic channel gain variances between different nodes in the network, was also assumed. If the random users' spatial distribution, and the associated propagation path losses between different nodes in the network, are taken into consideration, then these assumptions, in general, are no longer valid.

Moreover, it is of great importance for service providers to improve the coverage area in wireless networks without the cost of more infrastructure and under the same quality of service requirements. This poses challenges for deployment of wireless networks because of the difficult and unpredictable nature of wireless channels.

In this chapter, we address the relay assignment problem for implementing cooperative diversity protocols to extend the coverage area in wireless networks. We study the problem under the knowledge of the users’ spatial distribution which determines the channel statistics, as the variance of the channel gain between any two nodes is a function of the distance between these two nodes. We consider an uplink scenario where a set of users are trying to communicate to a base station (BS) or access point (AP) and describe practical algorithms for relay assignment.

In broadband wireless communications, the channel exhibits frequency selectivity (delay spread), resulting in inter-symbol interference (ISI) that can cause serious performance degradation. A mature technique to mitigate the frequency selectivity is to use orthogonal frequency division multiplexing (OFDM), which eliminates the need for high complexity equalization and offers high spectral efficiency. In order to combine the advantages of both the MIMO systems and the OFDM, space–frequency (SF)-coded MIMO-OFDM systems, where two-dimensional coding is applied to distribute channel symbols across space (transmit antennas) and frequency (OFDM tones) within one OFDM block, can be developed to exploit the available spatial and frequency diversity.

If longer decoding delay and higher decoding complexity are allowable, one may consider coding over several OFDM block periods, resulting in space–time–frequency codes to exploit all of the spatial, temporal, and frequency diversity.

The chapter is organized as follows. First, we focus on SF-coded MIMO-OFDM systems in broadband scenario and introduce two systematic approaches to design SF codes to achieve full spatial and frequency diversity within each OFDM block. Then, we consider STF coding for MIMO-OFDM systems, where the coding is applied across multiple OFDM blocks to exploit the spatial, temporal, and frequency diversity available in broadband MIMO wireless communications.

Space–frequency diversity and coding

In this section, we focus on SF-coded MIMO-OFDM systems to achieve the spatial and frequency diversity in broadband wireless communications. First, we specify an SF-coded MIMO-OFDM system model and discuss design criteria for achieving the full spatial and frequency diversity.

Routing is the process of transferring data packets from one terminal to another. Routing aims to find the optimal path according to some criterion. Shortest-path routing is a common scheme used for routing in data networks. It depends on assigning a length to each link in the network. A path made up of a series of links will have a path length equal to the sum of the lengths of the links in the route. Then, it chooses the path between source and destination that has the shortest route. The shortest-path route can be implemented using one of two well-known techniques, namely, the Bellman–Ford algorithm or the Dijkstra's algorithm [11].

In mobile ad hoc networks (MANETs), data packet transmissions between source and destination nodes are done through relaying the data packets by intermediate nodes. Hence, the source needs to locate the destination and set up a path to reach it. There are two types of routing algorithms in MANETs, namely, table-based and on-demand algorithms. In table-based routing algorithms, each node in the network stores a routing table, which indicates the geographic locations of each node in the network. These routing tables are updated periodically, through a special HELLO message sent by every node. Table-based routing protocols for MANETs include the destination sequence distance vector routing protocol (DSDV), wireless routing protocol (WRP), and cluster-head gateway switch routing (CGSR). The periodical updating of the routing tables makes table-based routing algorithms inefficient.

In static networks essentially all functionality resides in the network access stations (NASs). The performance of the network is therefore determined by how the NASs provide logical connectivity and throughput to satisfy the network's traffic requirements. This chapter explores the performance issues in static networks, viewing them all as special cases of shared media, as described in Section 5.1. Existing and potential uses of shared media abound, the most important of these being to provide efficient local access for end users to a larger optical network. The multiplexing and multiple-access techniques required to achieve multipoint logical connectivity in these networks are treated in Section 5.2. Sections 5.3 through 5.6 deal with capacity allocation and control to serve prescribed traffic requirements. We first point out some general flow conservation constraints that must be satisfied in any shared-channel system. Then the problems of traffic scheduling and control are discussed in settings with increasing degrees of complexity: dedicated connections (Section 5.4), demand-assigned connections (Section 5.5), and packet switching in the optical layer (Section 5.6). Section 5.7 discusses network access applications of static multipoint architectures. These include broadcast star-based and wavelength-router-based passive optical networks (PONs) that provide the foundation of fiber to the home/premises. In these applications the static network is the link between the end user and an optical core or metropolitan area network.

Shared Media: The Broadcast Star

The simplest form of a transparent optical network, the static network, was defined in Chapter 3 as a collection of fixed (passive) splitting/combining nodes without wavelength selectivity, interconnected by fibers that provide full or partial connectivity among a set of NASs.

Given a source s and destination d for a point-to-point connection on a selected waveband in an LLN, the Min-Int algorithm presented here attempts to find a minimum-interference optical path p = 〈s, d〉 for that connection on the given waveband. The exact sense in which interference is minimized requires some explanation and is defined in Section E.3.

The Image Network

The approach used to find a path that minimizes interference is based on shortest path calculations, where the path “length” takes into account weights or “lengths” representing currently active interfering signals. These weights are associated with nodes rather than links. A useful way of visualizing the node-weighting procedure is shown in the image network of Figure E.1. In the network shown in the figure, each node of the original network is “blown up” to create additional intranodal links between each input/output port pair. This is nothing more than a representation of the internal structure of the LDC on the chosen waveband (see Figure 2.19[b]). The image network of Figure E.1 corresponds to the state of activity in the network of Figure 6.55. Two optical connections, (1, 1*) and (2, 2*), are active, with signal S1 transmitted from station 1 to 1* and signal S2 transmitted from station 2 to 2*.

We shall denote an internodal link from node i to node j by (i, j) and assign it a positive weight d(i, j).

The concept of the limiting cut, introduced in Section 6.3.1.2, stems from the Min Cut–Max Flow relation in multicommodity flow problems. We first give a brief summary of this problem and then present a heuristic for finding limiting cuts.

The Multicommodity Flow Problem and Limiting Cuts

In the most common version of the multicommodity flow problem, a set of demands are prescribed between source-destination node pairs in a network with a given topology and link capacities. (Each source-destination demand is known as a commodity, and the network can be anything—gas pipelines, airline routes, highways, and so on.) The basic issue is whether the prescribed demands can be satisfied within the capacity constraints; that is, whether all commodities can be routed through the network (in a bifurcated manner if necessary) so that the total flow of all commodities on each link does not exceed its capacity. If so, the demands are said to be feasible.

In wavelength-routed networks (WRNs), the commodities (demands) are LCs, each requiring one λ-channel, so the capacity of a cut Ci is FiW, where Fi is the number of fiber pairs in the cut and W is the number of available wavelengths. Because a channel in a WRN is a single point-to-point entity, bifurcated routing is not permitted in a WRN. (An exception would be a case in which several λ-channels are required to carry the flow on one LC.)

The relations between cut capacities and feasible demands were stated in Section A.1.8 for the single-commodity case. In the multicommodity case, which is of interest here, the relations are considerably more complex.

SONET, the acronym for synchronous optical network, is currently the prevailing standard for high-speed digital transmission in North America. Introduced in the 1980s, it replaced an earlier standard, the plesiochronous digital hierarchy (PDH), which had been in place for more than two decades prior to the introduction of the SONET standard [Ballart+89]. The most frequently used lower levels of the PDH system are the DS1 (1.544 Mbps, designed to carry 24 64-Kbps digitized voice signals plus synchronizing overhead) and DS3, running at 44.736 Mbps. An architecture similar to SONET, the synchronous digital hierarchy (SDH), is currently used in Europe and Japan, replacing an earlier European hierarchy similar to the PDH system. SONET can carry PDH bit streams as well as many other types of digital traffic (e.g., ATM cells) as part of its payload. One of the most important features of SONET is its highly organized protection capability [Wu92].

The basic building block (i.e., the first level) of the SONET signal hierarchy is called the synchronous transport signal-level 1 (STS-1). STS-1 has a bit rate of 51.84 Mbps and is divided into two portions, transport overhead and information payload, and the transport overhead is divided further into line and section overheads. (A line is composed of one or more sections in series, separated by electronic regenerators.) The line overhead is terminated at SONET terminals and add/drop multiplexers (ADMs), and the section overhead is terminated at regenerators.

Survivability against failures, including failure recovery, is important in any telecommunications network but is highly critical for high-bandwidth optical networks. As more traffic is concentrated on fewer routes, the number of customers that can be potentially affected by a failure is increased. An analysis of failures in the Public Switched Telephone Network over a two-year period in the 1990s showed that human error, acts of nature, and overloads were the major sources of failure. The impact of the failures was measured in terms of how many times a particular failure occurred, duration of the outage, and number of customers and number of customer minutes affected during that outage. During that period, the average number of customers affected due to cable cuts or cable component failures was 216,690, costing 2,643 million in customer minutes. Similarly, the average number of customers affected by each equipment failure was 1,836,910, costing 3,544.3 million in customer minutes . Cable cuts and hardware/equipment failures account for approximately half of the failures encountered in the network during that period.

Fiber cuts are considered one of the most common failures in fiber-optic networks. Furthermore, the use of WDM over these fibers produces an extremely high volume of traffic on a cable. Commercially available fiber-optic transmission systems can run at 10 Gbps or more per channel with 80 or more channels (wavelengths) per fiber. This translates to more than 800 Gbps per fiber.