This work was supported by the National Basic Research Program of China (“973” Program) under Grant No. 2009CB320404, National High Technology Research and Development Program of China (“863” Program) under Grant No. 2007AA01Z217, and National Natural Science Foundation of China under Grant No. 60972048.
In future mobile communication system, the advantage of Multiple-Input Multiple-Output (MIMO) technology will be widely accepted. Restricted by node size and energy, the implementation of MIMO technology is difficult[1,2]. Cooperative communication technology makes full use of the all-direction transmission feature of radio signals. Nodes in the radio network cooperate with each other to form a virtual antenna array to obtain the spatial diversity gain of the MIMO technology.
Main modes of cooperative communication include: Amplify and Forward (AF), Decode and Forward (DF), and Coded Cooperation (CC). Compared with other cooperative modes, coded cooperation integrates cooperative communication with channel code technology. In this mode, no more system resources are utilized to obtain complete diversity gain. In the cooperative system, it is vital to select an appropriate cooperative node. It even determines whether the cooperative system can bring gains. Many researchers are contributing to the study of this issue.
1 Evaluation Standards for Relay Node Selection Algorithm
We evaluate the performance of the relay selection algorithm from the following aspects:
(1) Algorithm Effects
The major goal of the cooperative system is to increase the network capacity, reduce power consumption, and expand network coverage. Naturally, the goal becomes one of an appraisal standard of the cooperative node selection algorithm. The tradeoff is in network capacity, power consumption, and network coverage. Therefore, the cooperative node selection algorithm should choose different optimization target according to different system requirements.
(2) Algorithm Complexity
The essence of cooperative communication is to optimize the whole system from a network perspective. But this introduces more optimization elements which cause an increase in algorithm complexity. It is an important standard for evaluating the cooperative node selection algorithm to control the algorithm complexity of cooperative node selection in order to achieve better system performance.
(3) Communication Overhead Caused by Algorithm
In the cooperative system, more information should be transmitted. As a result, the communication overhead is increased, which impacts the system negatively. Therefore, it should be taken into consideration in the cooperative node selection algorithm. Cooperation should be selected only when the cooperative gain is greater than the performance loss of extra overhead. In the implementation of cooperative node selection algorithm, overhead should be reduced. The main methods include limited feedback and fuzzy selection.
(4) Algorithm Auto-Adaptation and Error-Tolerance
Owing to the time-variation and node mobility, channel information and node state information cannot be obtained accurately. As a result, the cooperative node selection algorithm should be robust and able to adjust the selection policy in an auto-adaptation mode. At the same time, it should be error-tolerant of the worst channel environment and no-response of the cooperative node.
(5) Algorithm Price (Supported by Software and Hardware)
Evaluation of the algorithm covers the time price and the space price. Different cooperative node selection algorithms require different software and hardware. Cooperative node selection algorithm should be selected according to the application background, demand, technology, and cost.
2 Category of Relay Node Selection Algorithm
(1) Implementation Mode of the Algorithm
The implementation mode of the algorithm includes: central mode and distributed mode. The central algorithm means that the required information is transmitted to a central node (such as base station and AP), which in turn utilizes the information to implement the cooperative node selection algorithm and return the result to the source node and the corresponding cooperative node. The distributed algorithm implies that, according to the information exchange and coordination, the node determines whether to cooperate and who to cooperate with.
The advantage of the central algorithm is that the planning is based on a global prospective. Consequently, the system is working in a globally optimal state. Owing to the collection of relevant information and calculation of a global optimum, major communication overhead and calculation overhead are generated. The distributed algorithm is usually only partly optimal. But the distributed algorithm disperses communication overhead and calculation complexity. In addition, the distributed algorithm is more applicable to networks without fixed support (such as Ad Hoc network).
(2) Number of Relay Nodes
Determining the number of relay nodes is a primary concern of the relay node selection algorithm. Whether to use a single node or use multiple nodes remains an open question. To use a single cooperative node, the hardware at the receiving end is simple and easy to implement. In addition, the diversity steps are not lost. Single relay node selection requires the information of each channel, and the information should be sorted before the optimum node is selected. The processing capability and supported power of a single node are limited. When the channel is in deep recession, a single relay node cannot implement the QoS requirements of the source node. In addition, multiple relay nodes can increase the multiplexing gain of the system. Therefore, the selection algorithm, which adjusts the count of node selection according to the channel and relay node states, is more reasonable.
(3) Cooperation Modes
Different cooperation modes significantly impact the selection algorithm of the cooperative node. For example, in the DF cooperation mode, the properly decoded node can participate in cooperative transmission. In the AF, the cooperative node does not process the source node signals and all cooperative nodes can transmit the information. It affects the alternative collection of the cooperative node selection algorithm. Therefore, different cooperative node selection algorithms should be selected for different cooperative modes. In addition, we can integrate cooperation mode selection with cooperative node selection. In the same system, different cooperation modes, and cooperative node selection algorithms in an adaptive mode can be used.
(4) Joint Allocation of Relay Node Selection and Other Cooperative Resources
For the cooperative system, the cooperative node is only one part of the system resources. Therefore, the current research takes cooperative node selection and other resource allocation such as power and bandwidth into consideration. Through cross layer design, system resources can improve the system performance. But owing to the introduction of more variables and optimization goals, system design is faced with a great challenge. In most cases, the system optimization problem becomes a Non-Polynomial (NP) problem. How to find the appropriate joint optimization parameters and design executable progressive optimum algorithm is the key to cooperative node selection and other resource allocation algorithm.
(5) Application Scenario
The current wireless communication system can be divided into infrastructure and non-infrastructure. In infrastructure networks (such as cell networks), the communication mode is multiple-to-one or one-to-multiple communication; namely, multiple users to base station or base station to multiple users. In addition, the central node of the network manages and dominates the network, which is helpful for the reasonable allocation of resources and the implementation of a central algorithm. In non-infrastructure networks (such as Ad Hoc networks), multiple source and destination node pairs exist. There is no central node managing the network. The communication node pairs are competitive. Therefore, controlling the inter-interference between communication node pairs is a key factor affecting system performance and is also a design challenge.
(6) Relay Node Attributes
In different networks, the attributes of relay nodes are different. Relay nodes can be fixed or mobile, can be active or inactive. Some nodes are equipped with a single antenna and some are equipped with multiple antennas. The different node attributes affect the policy of cooperative node selection. In the cell network, mobile or fixed relay nodes are supported by energy. In addition, in most relay nodes, multiple antennas can be equipped, to possess powerful processing and transmission capabilities. Therefore, much work can be transmitted to the relay node to lower the complexity and energy consumption of mobile terminals. At the same time, better QoS can be provided for mobile terminals. In self-organized networks, the attributes of all nodes are basically the same and most operate on battery power. Thus, the processing capability is limited. As a result, in the design of cooperative node selection algorithm, the energy issue should be taken into consideration. The lifetime of the network can be expanded on the precondition that the service is guaranteed.
3 Typical Algorithms
(1) Single Node Selection Based on Cooperative Gain
The purpose of the cooperative node selection algorithm is to increase the gains of cooperative communication. Reference  studies the cooperative node selection policy in coded cooperation. The model of the cooperative network is shown in Figure 1. This paper takes the end-to-end Frame Error Rate (FER) as the standard to define the user cooperative gains G :
In Equation 1, the Pno-coop is the FER of non-cooperative transmission. Pcoop is the FER of cooperative transmission. Therefore, only when G >1, can cooperation be used. As a result, the cooperative region with gain is obtained. At the same time, the cooperative node selection standard is provided. It selects the cooperative node which can bring the maximum cooperative gain to participate in cooperation. The algorithm requires the support of node location information, thus extra hardware devices (such as GPS) or localization algorithms should be used.
(2) Distributed Relay Selection Based on instantaneous channel State
Owing to the time-variation feature of the wireless channel, an adaptation feature is required in cooperative node selection algorithms. Reference  provides a distributed cooperative node selection algorithm. The algorithm works with traditional 802.11 protocol, utilizes Request-To-Send (RTS) and Clear-To-Send (CTS) to evaluate the channel state between the source node and relay node and between the destination node and the relay node. After a relay node obtains the channel information, it makes a judgment. The standard of the judgment is as shown in Equation 2.
Then, a backoff timer is started. The backoff time T is inversely proportional to the channel condition hi. The relay node with better channel condition will be given priority when entering the channel. It sends a flag frame to notify the source node, destination node and other relay nodes. Then, a relay selection is complete, namely, the best cooperative node from M relay node collection is selected. The algorithm selects a relay node according to the instantaneous channel condition. With the change in channel fading, different relay nodes are selected at different time, as shown in Figure 2. But in this algorithm collision may occur at the node selection stage and consequently the relay node cannot be selected properly.
Figure 2. Dynamical selection of optimum relay node.
(3) Cooperative Node Selection Based on Cluster
Reference  studies the cooperative node selection when multiple pairs of source and destination nodes exist. To attain the diversity gain across the entire network, the node selection protocol should provide enough cooperative nodes for each sending node. Then, the nodes form many groups, as shown in Figure 3.The cooperative nodes in each group can properly decode the information of the sending node with high probability.
In this paper, for the distributed scenario, a simple static cooperative node selection policy is provided. The policy ensures that all sending nodes in the network can obtain n+1 diversity gain. N is the number of cooperative nodes. First, each node maintains a preferred cooperative table. The first n nodes in the table have priority. There are many design and implementation methods for the preferred cooperative table. One simple implementation method is:
[i +1,i +2,…,M,1,2,…,i -1], i is the ID of the current node.
In the central control scenario, the central control node has the channel information. Therefore, the best solution from all possible results can be selected to obtain extra performance gain.
Figure 4 compares the distributed and central algorithms.
(4) Cooperative Node Selection Based on Energy
In wireless networks such as cell networks, Wireless Local Area Network (WLAN), and wireless sensor network, most terminals use battery power supplysupplfor power supply. As a result, the usable resources in the network are limited. Therefore, how to use the network resources efficiently and how to extend the lifetime of the network is an important issue.
With the emergence of cooperative communication technology, the resources (channel and energy) of the cooperative nodes are shared. It provides an effective way of saving node energy and expanding the lifetime of the network. Recent contributions to the expansion of network life through the cooperation technology has been made by some researchers.
In Reference , the author, based on the model in Figure 5, takes the Channel State Information (CSI) and Rest Energy Information (REI) into consideration and selects the best cooperative node to complete cooperation transmission. As a result, the lifetime of the AF cooperative network is expanded. In Reference , the author makes the algorithm more practicable in the same model through the adjustment range of the discrete power. At the same time, the Markov chain is used to evaluate the network lifespan.
Similarly, in Reference , in the DF cooperative network, node power consumption is reduced and the life of the network is extended through the inspired relay node selection, power allocation method, and layout of relay nodes. In Reference , the transmission distance is increased through cooperative beamforming. As a result, nodes with low node energy are prevented, thereby expanding the lifetime of the network.
(5) Cooperative Node Selection Based on Cross-Layer Optimization
Joint consideration of cooperation node and other system resource allocation can improve system performance significantly.
Reference  studies the cell network possessing one base station and multiple mobile stations as shown in Figure 6. The access mode of the network is Orthogonal Frequency Division Multiple Access (OFDMA). In the document, the central optimization model is created and the multi-level optimization parameters are taken into consideration. Finally, optimum power and bandwidth allocation are obtained, while the best cooperative node selection mode and cooperative policy selection method are also obtained.
In addition, Reference  obtains the distributed optimization method through the Lagrangian dual decomposition. It also considers the impact of flow control.
(6) Selection Policy Based on the Multiple Cooperative Nodes
In wireless networks, through the cooperation of multiple nodes, spatial diversity gain and spatial multiplexing gain can be obtained. At the same time, the service load can be balanced. Reference  assumes the destination node has multiple antennas, and uses different cooperative nodes to transmit different information to obtain spatial multiplexing. The diversity gain is obtained through the cooperative node selection, as shown in Figure 7. The document searches for the best cooperative node through the greedy search. According to the analysis, better diversity and multiplexing compromising can be obtained using multiple nodes to implement cooperation.
In Reference , two relay node selection methods are presented. In the fixed selection policy, the source node selects M relay nodes from K cooperative nodes. The M relay nodes send information received from the source node at the same time. The document also put forward a threshold-based dynamic relay selection method. It can minimize the number of cooperative nodes participating in cooperation thereby reducing the probability of interruption. As a result, the interference caused by cooperation is reduced.
In Reference , when multiple relay nodes exist, power control with node selection can be combined. The purpose of the protocol is to allocate the power of the source node and relay node set to minimize the probability of interruption and reduce the complexity of calculation. The protocol covers two stages: selecting the power of the source node after weighing the S-R and R-D channel conditions, and selecting the relay node set that can minimize interruption through search. Power in the relay nodes of the set is equally allocated. According to the emulation and analysis, the algorithm can reduce the probability of interruption.
Generally, the idea of cross-layer design for dynamically allocating system resources makes full use of cooperative communication. Compared with traditional systems, major performance gain is obtained. But dynamic resource allocation introduces more variables, making system optimization harder and thus brings a big challenge for the implementation. The cooperative communication system should be optimized by selecting proper optimization parameters and controlling complexity.
 NOSRATINIA A, HUNTER T E, HEDAYAT A. Cooperative communication in wireless networks [J]. IEEE Communications Magazine, 2004,42(10): 68-73.
 LIU Pei, TAO Zhifeng, LIN Zinan,et al. Cooperative wireless communications: A cross-layer approach [J]. IEEE Wireless Communications, 2006,13(4): 84-92.
 LANEMAN J N, TSE D N C, WORNELL G W. Cooperative diversity in wireless networks: Efficient protocols and outage behavior[J]. IEEE Transactions on Information Theory, 2004,50(12): 3062-3080.
 HUNTER T, NOSRATINIA A. Diversity through coded cooperation[J]. IEEE Transactions on Wireless Communications, 2006,5(2): 283-289.
 LIN Z, ERKIP E, STEFANOV A. Cooperative regions and partner choice in coded cooperative systems[J]. IEEE Transactions on Communications, 2006, 54(7): 1323-1334.
 BLETSAS A, REED D P, LIPPMAN A. A simple cooperative diversity method based on network path selection[J]. IEEE Journal on Selected Areas in Communications, 2006,24(3): 659-672.
 NOSRATINIA A, HUNTER T E. Grouping and partner selection in cooperative wireless networks[J]. IEEE Journal on Selected Areas in Communications, 2007,25(2):369-378.
 HUANG Wan-Jen, HONG Yao-Win, KUO C C J. Lifetime maximization for amplify-and- forward cooperative networks[C]//Proceedings of Wireless Communications and Networking Conference(WCNC'07), Mar 11-15,2007, Hong Kong, China. New York, NY,USA: IEEE,2007: 814-818.
 HUANG Wan-Jen, HONG Yao-Win, KUO C C J. Discrete power allocation for lifetime maximization in cooperative networks[C]//Proceedings of the 66th Vehicular Technology Conference(VTC-Fall’07), Sep 30-Oct 3,Baltimore, MD,USA. Piscataway, NJ,USA: IEEE, 2007: 581-585.
 HIMSOON T, SIRIWONGPAIRAT W P, HAN Zhu, et al.Lifetime maximization via cooperative nodes and relay deployment in wireless networks[J]. IEEE Journal on Selected Areas in Communications, 2007,25(2): 306-317.
 HAN Zhu, POOR H V. Lifetime improvement of wireless sensor networks by collaborative beamforming and cooperative transmission[C]//Proceedings of IEEE International Conference on Communications (ICC’07), Jun 24-28, 2007, Glasgow, UK. Piscataway, NJ,USA: IEEE, 2007: 3954-3958.
NG T, YU W. Joint optimization of relay strategies and resource allocation in cooperative cellular networks[J]. IEEE Journal on Selected Areas in Communications, 2007,25(2): 328- 229.
 LE L, HOSSAIN E. Cross-layer optimization frameworks for multihop wireless networks using cooperative diversity[J]. IEEE Transactions on Wireless Communications, 2008,7(7): 2592-2602.
 ZHANG Shunqing,LAU V K N.Design and analysis of multi-relay selection for cooperative spatial multiplexing[C]//Proceedings of IEEE International Conference on Communications (ICC’08), May 19-23,2008, Beijing, China. Piscataway, NJ,USA: IEEE, 2008: 1129-1133.
 BAN Tae Won,JUNG Bang Chul,SUNG Dan Keun,et al. Performance analysis of two relay selection schemes for cooperative diversity[C]//Proceeding of IEEE 18th International Symposium on Personal, Indoor and Mobile Radio Communications(PIMRC’07), Sep 3-7,2007, Athens, Greece. Piscataway, NJ,USA: IEEE,2007: 5p.
 KIM J, HAN M, HAN S, et al. A new power allocation scheme with relay selection in multinode AF relay networks[C]//Proceedings of IEEE Military Communications Conference(Milcom’07), Oct 29-31,2007, Orlando,FL, USA. Piscataway, NJ,USA: IEEE, 2007: 5p.