To the best of our knowledge, there is no work considering the same scenario which consists of multiple users, multiple relays and only a single destination in bidirectional relay network as ours. Nevertheless, the proposed scenario is meaningful. Regarding the multi-source nodes and the destination as mobile handsets and base station, respectively, this scenario is more similar to the communication system in reality compared with those consisting of only one source and one destination. On the other hand, we take multiuser interference into account, which is not considered in the most research work related to multiuser two-way relaying networks. Channel orthogonality is assumed frequently in many studies to avoid interference. However, such assumption is not so realistic and de-grades the bandwidth efficiency. To be closer to reality or more bandwidth efficient, taking the multiuser interference into consideration is necessary. In our work, we consider a code division multiple access (CDMA) system with nonorthogonal spreading sequences.Signal to interference-plus-noise ratio is a benchmark of communication quality. For the sakes of increasing SINR as well as facilitating the implementation, relay selection based on maximizing the SINR of worse link is performed in our work. Besides, aiming at miti-gating the interference, we consider the design of linear filter at each relay as well. The result shows that the linear filter is similar to the minimum mean-squared error detector.
Furthermore, we simulate the proposed scheme with several different parameters such as the number of users and relays, and the length of spreading sequences. Also, we compare the proposed relay selection method with random relay selection approach, and
the result shows that our proposed method has better performance in terms of the bit error rate.
To sum up, the contributions of research include:
• We develop a more realistic scenario in bidirectional relay network consisting of multiple sources, multiple relays, and one destination.
• Relaxing the constraint of channel orthogonality, we perform relay selection while taking the multiuser interference into account. Based on the selection criterion that aims at maximizing the signal to interference-plus-noise ratio of the worse link, our proposed scheme outperforms the random relay selection method in terms of the bit error rate.
• The multiuser interference is mitigated by the linear filtering at each relay, which is designed by maximizing the SINR of the worse link as well.
Chapter 2
Background Review
2.1 Relay Networks
In wireless communication, several challenges such as limited energy, service coverage and channel impairments caused by multi-path propagation and Doppler shifts should be overcame. Relay networks have been proposed to conquer these difficulties by exploiting the spatial diversity gains without the need of multiple antennas at each node [8] [12]. The basic concepts of relay networks were first introduced in [13]. Cover and El Gamal stud-ied a network which consisted of a source, a destination and a single relay. The focus of this work was on evaluating the channel capacities of the Gaussian relay channel and cer-tain discrete relay channels based on the information theoretic properties. Different from recent works, the analysis of capacity in [13] was an additive Gaussian channel noise, however, fading channel is considered in most recent works [1]. The fundamental trans-mission process of a relay network is as follows. At first time instant, the source sends its information to the relay. The relay then processes the received signals and forwards them to the destination. After properly combining (e.g., maximum ratio combining) the signals sent from the source and relay at the destination, the advantages of the relay com-munication such as spatial diversity can be achieved. In short, the relay comcom-munication is recognized as an effective method to attain broader coverage range and mitigate channel impairments due to fading.
Depending on the number of information flows, there exist two different network
configurations: one-way relay network and two-way relay network. In one-way relay net-works, information is transmitted in a single direction from the source to the destination.
However, in two-way relay networks, the information is allowed to transmit in opposite directions such that the two sources can exchange information with the aid of relays. In the following, the concepts of one-way and two-way relay networks will be discussed.
2.1.1 One-Way Relay Networks
The fundamental concept of one-way relay network is that the information can only be transmitted in one direction, i.e., from a source to a destination. A typical one-way relay network consisting of a source, a destination and a single relay is depicted in Fig. 2.1. In Fig. 2.1, the communication is established in two time slots with the aid of the relay node R. In the first time slot, the source node S broadcasts its symbol to the relay and the in-tended destination. Upon receiving the signal, the relay processes it based on some kinds of relaying strategies to regenerate a new signal. After that, the relay retransmits the new signal to the destination terminal D in the second time slot to complete the information transmission.
Many relaying strategies have been proposed for relays to execute on their received signals. Some of these techniques are amplify-and-forward (AF), decode-and-forward (DF), compress-and-forward (CF), and estimate-and-forward (EF). Among these approaches, AF and DF are the most well-known ones. In the AF scheme, each relay simply amplifies its received signal and retransmits the amplified signal to the destination, whereas in the DF scheme, each relay should detect the received signal and retransmits the detected sig-nal. Following is an example to demonstrate the regenerated signals at relay nodes in AF and DF schemes.
Example [12]
Consider the system model depicted in Fig. 2.1, but the direct link (i.e., S-D link) does not exist for simplicity. In the first time interval, source S transmits its symbol xsto the relay node R. The received signal at the relay can be expressed as
yr = hxs+ n (2.1)
Figure 2.1: A one-way relay network: source S transmits its information to the destination D with the help of relay.
where h is the channel coefficient for S-R link, and n denotes the additive noise. In the second time interval, the relay regenerates a new signal xr = f (yr) and transmits it to the destination. The function f (·) stands for different relaying strategies. In AF scheme, xr can be presented as
xr = βyr (2.2)
where β is an amplified coefficient. However, it is not the case for DF scheme. If DF strategy is employed, xr can be shown as
xr = ˆxs (2.3)
where ˆxsis the decoded symbol of xs.
All in all, the AF approach is more simpler than the DF method because signal de-tection does not be needed at each relay in the AF approach. Therefore, the DF method requires more processing power at the relays compared to the AF approach. There have been a lot of works done on AF and DF relaying schemes. Although different strate-gies are performed at relays, spatial diversity can be achieved for both relaying schemes because of the reason that independent replicas of the source signal are received by the destination if the direct link exists.
Figure 2.2: A two-way relay network: sources S1and S2exchange information with each other with the aid of the relay. (a) Traditional scheme. (b) TDBC scheme. (c) MABC scheme.
2.1.2 Two-Way Relay Networks
In many applications of relay networks, two end-sources may need to exchange informa-tion with the aid of relays. The concept of two-way channels can be traced back to the work of C. E. Shannon in 1961. Shannon obtained an inner bound and an outer bound to the rate region for a full-duplex scenario. However, no relay node existed in Shannon’s work at that time. When the distance between two end-sources is so long that the direct link between them is not available or when the channel quality between two end-sources is poor, the communication between two end-sources is unreliable. Assisted by relays, a more trustworthy communication can be established between two end-sources.
Since two-way relay network is more bandwidth efficient than one-way relay network, it has received considerable attention recently. Several two-way relay network protocols have been proposed: the traditional technique, the time division broadcast (TDBC) proto-col and the multiple access broadcast schemes (MABC). A typical two-way relay network consisting of two end-sources and one relay is depicted in Fig. 2.2. As shown in Fig. 2.2 (a), a traditional two-way relay network requires four time slots to accomplish the infor-mation exchange between the two end-sources. In the first time slot, source S1 broadcasts its symbol to the relay. Then, the relay retransmits a new signal to source S2in the second time slot after performing some kinds of relaying strategies at the received signal. In the third and fourth time slots, the same procedures in the first two time slots are conducted again. However, the information flow is from source S2to source S1. Consequently, this
traditional scheme is not bandwidth efficient. As shown in Fig. 2.2 (b), the TDBC proto-col based on the concept of network coding reduces the number of time slots to three. In the first two time slots, sources S1 and S2transmit their symbols to the relay sequentially.
It is worth noting that the relay has to decode the received symbols and perform an XOR operation on the decoded signals before retransmitting a new signal to sources S1 and S2. In other words, if the transmitted symbols by sources S1 and S2 are xs1 and xs2, then the regenerated signal at the relay can be expressed as xr = ˆxs1 ⊕ ˆxs2, where ˆxs1 and ˆxs2 denote the decoded symbols of xs1 and xs2, respectively. As a result, each source can retrieve its desired signal easily by performing an XOR operation on the received signal and its transmitted signal. Since the concept of network coding is used, the TDBC scheme provides a throughput which is significantly higher than the traditional relaying scheme.
The MABC schemes are shown in Fig. 2.2 (c). There are two well-known protocols in the MABC schemes: the analog network coding (ANC) [14] and the physical-layer network coding (PNC) [5] [15]. For both protocols, two time slots are required to ac-complish the information exchange between the two end-sources. In the first time slot, the two end-sources transmit their signals to the relay simultaneously. In the second time slot, the relay retransmits the mixed version of two incoming signals. Compared with the TDBC protocol, the MABC schemes have better bandwidth efficiency. However, under a half-duplex constraint, the MABC schemes can not utilize the direct link between two end-sources even if the link exists. To sum up, the MABC schemes are more bandwidth efficient while the TDBC protocol can offer more reliable communication quality than the MABC schemes because of the utilization of the direct link. For example, in Fig. 2.2, the diversity order of the MABC schemes is one while that of the TDBC protocol is two under a half-duplex constraint.