hop neighboring nodes are found. Then initial probability of channel occupancy will be generated. Initial probability will be generated for the nodes present in the network. A uniform number generator is used for this process. The nodes with 0 probability will be neglected because it is not using any channel. Then the probability will be set as 0 if the probability is less than 0.6, and if it is greater than 0.6 then it will be set as 1. The node having more probability in number of occurrence will be detected as selfish node or attacker. The base station monitors this
Figure 3. Flowchart of proposed ICOOPON.
channel allocation process. When base station detects, that one or more number of nodes are having more occurrences of high probability of channel occupancy, then those nodes will be declared as selfish nodes.
In HMM, until the end state is reached the sequence of state is generated by moving from state to state according to the state transition probabilities. In the HMM, the state themselves are not directly observed instead we assume that there are some sequence of state in Markov chain that cannot be observed directly, and these states generate observations. Each state may have different distribution. The HMM model falls in a subclass of Bayesian network.
5. Simulation Result and Analysis
5.1. Network Scenario
The selfish attacks have been simulated in NS-2. The wireless scenario of 1000 × 1000 square meter area has been considered. Number of nodes has been varied from 10 to 100 at the interval of 20. The simulation results in the presence of selfish nodes and in the absence of selfish nodes has been presented. The parameters that have been analyzed are packet delivery ratio, throughput and end to end delay. In simulation scenario maximum 20 nodes are considered as selfish nodes.
Number of selfish nodes is varied from 2 to 20 and performance of COOPON and ICOOPON is observed. For communication all the nodes use Omni-directional antenna (Table 1).
5.2. Simulation Results
The parameters such as Throughput and Packet delivery ratio has been considered for evaluating the performance of the algorithm.
1) Throughput: It is measured in data packets per second. It shows the rate of information or message delivered over a communication channel.
2) Packet Delivery Ratio: It shows the ratio of number of packet transmitted to number of packet that has been received.
3) End to end delay: I t is the total delay that a packet experience as it is travelling through a network.
The following Figure 4 shows a network scenario with 50 secondary users. The nodes shown in square block are selfish secondary users, nodes 9 and 11 are the selfish secondary users that are marked in red and shown as Selfish Secondary User. The source nodes are shown in red circles and destination nodes are marked as blue. The nodes with ID 1, 3, 6 etc. are the source nodes and nodes with ID 2, 4, 7 are the destination
Table 1. Simulation details.
Figure 4. Output of the COOPON with 50 secondary users.
nodes that are marked in blue. The nodes are continuously move and packet transmission takes place among the nodes.
Figure 5 depicts a scenario for 100 Secondary Users with 2 selfish nodes. Where nodes 9, 11 are selfish nodes that are marked as red. The nodes are the source nodes 1, 3, 6 and the destination nodes are with ID 2, 4 and 7. The nodes are continuously move and packet transmission takes place among the nodes. Simulations have been performed with 50 and 100 numbers of secondary users with varying selfish node attacks.
Figure 6 shows the average throughput ratio of AODV, COOPON and ICOOPON, which is the rate of successful message delivery over a communication channel. Here, AODV is included in all the plots, just to show how AODV is affected due to selfish attacks. It is measured in bits/second or data packets/second. Here, ICOOPON shows more throughput performance than the COOPON and AODV protocol technique. When the number of selfish node increases the throughput performance of COOPON and AODV technique will undergo degraded performance. Since ICOOPON works with HMM algorithm along with the COOPON technique, the performance of ICOOPON is improved.
Figure 7 shows the performance of packet delivery ratio in AODV, COOPON and ICOOPON protocols respectively. The packet delivery ratio shows the ratio of number of packets transmitted to number of packets received. It should be as high as possible.
Figure 5. COOPON with 100 secondary users.
Figure 6. Average throughput.
Figure 7. Performance comparison of PDR with 50 secondary users.
From the graph the performance of packet delivery ratio is more for ICOOPON as compared to COOPON and AODV protocol technique. As, ICOOPON technique uses two algorithms, it checks with the probability condition with all the nodes present over there. The main drawback of COOPON technique is that it does not detect selfish attack to search beyond one hop network because of this reason the efficiency is less compared with ICOOPON technique.
Figure 8 shows plot of an end to end delay with varying selfish nodes from 0 to 20. As the number of selfish nodes increases the end to end delay also increases. For example the End to end delay is largest for AODV as compared to COOPON and ICOOPON. The COOPON technique has delay of 0.85 whereas for ICOOPON it is 0.6. Thus even if the number of attackers is increasing, the respective end to end delay is also increasing. However, there is less effect on ICOOPON as seen in the above figure.
Figure 9 shows the performance of End to End Delay for 50 secondary users with varying selfish nodes from 0 to 20. As the number of secondary users increases, performance decreases with increase in number of selfish nodes.
The following table summarizes the performance study of AODV, COOPON, and ICOOPON.
Table 2 gives the details of Packet Delivery Ratio for 50 and 100 numbers of secondary users. The performance of AODV, COOPON and ICOOPON been analyzed in the presence of 5 selfish attackers. In the case of 50 secondary users AODV is having the least packet delivery ratio of 0.69. The proposed ICOOPON technique has the highest packet delivery ratio of 0.92 for 50 secondary users. In the case of 100 secondary users AODV is having the least Packet delivery ratio i.e. 0.59. The ICOOPON approach is having the highest Packet delivery ratio of 0.82.
Figure 8. End to end delay of 100 secondary users with AODV, COOPON and ICOOPON.
Figure 9. End to end delay of 50 secondary users, with AODV, COOPON and ICOOPON.
Table 3 gives the details of average throughput ratio for 50 and 100 number of secondary users. The performance of AODV, COOPON and ICOOPON has been analyzed in the presence of 5 selfish attackers. In the case of 50 secondary users AODV is having the least average throughput ratio of 11.37 kbps. The proposed ICOOPON technique has the highest packet delivery ratio of 13.5 kbps for 50 secondary users. When the number of secondary users are 100, AODV is having the least value i.e. 9.58 kbps. The ICOOPON approach is having the highest value i.e. 11.38 kbps.
Table 4 gives the details of end to end delay for 50 and 100 numbers of secondary users. The performance of AODV, COOPON and ICOOPON has been analyzed in the presence of 4 selfish attackers. In the case of 50 secondary users the proposed ICOOPON technique has the least delay of 0.1884 sec and AODV has the highest end to end delay of 0.2637 sec. In the case of 100 secondary users ICOOPON approach has the least delay of 0.2 sec, and AODV is has the highest delay of 0.27 sec.
Table 2. Comparative study of AODV, COOPON and ICOOPON w.r.t. packet delivery ratio.
Table 3. Comparative study of AODV, COOPON and ICOOPON w.r.t throughput.
Table 4. Comparative study of end to end delay for AODV, COOPON, and ICOOPON.
From the above tables it can be summarized that when number of nodes increases the efficiency decreases for both the case of packet delivery Ratio, Throughput and end to end delay. However, the ICOOPON is least affected with increase in selfish node attacks as compared to COOPON and ICOOPON.
The cognitive radio is currently attracting numerous research efforts, where the major problem is of security. The novel technique, namely ICOOPON is proposed in this paper which has improved performance as compared to COOPON and AODV. For simulation 50 and 100 numbers of nodes have been used with four and five numbers of selfish nodes. The ICOOPON technique provides the improved performance due to HMM. The proposed technique is proved to be more efficient by 32% than the COOPON technique in various parameter performances such as packet delivery ratio, end to end delay and throughput. For future work cryptographic model and game theory can be applied to check the selfish attack detection analysis. The existing parameters can be taken into account for performance analysis and can be analyzed using jitter.
 Jo, M., Han, L., Kim, D. and In, H.P. (2013) Selfish Attacks and Detection in Cognitive Radio Ad-Hoc Networks. Proceeding of IEEE Network, 27, 46-50.
 Ding, L., et al. (2010) Cross-Layer Routing and Dynamic Spectrum Allocation in Cognitive Radio Ad Hoc Networks. IEEE Transactions on Vehicular Technology, 59, 1969-1979.
 Wang, W.F. (2009) Spectrum Sensing for Cognitive Radio. 2009 Third International Symposium on Intelligent Information Technology Application Workshops, 21-22 November 2009.
 Akyildiz, I.F., et al. (2008) A Survey on Spectrum Management in Cognitive Radio Networks. IEEE Communications Magazine, 46, 40-48.
 Kim, H. and Shin, K.G. (2008) Efficient Discovery of Spectrum Opportunities with MAC-Layer Sensing in Cognitive Radio Networks. IEEE Transactions on Mobile Computing, 7, 533-545.
 Akyildiz, I.F., Lee, W.-Y., Vuran, M.C. and Mohanty, S. (2008) A Survey of Spectrum Management in Cognitive Radio Networks. Georgia Institute of Technology, IEEE Communications Magazine.
 Cheng, G., et al. (2007) Joint On-Demand Routing and Spectrum Assignment in Cognitive Radio Networks. 2007 IEEE International Conference on Communications IEEE, 24-28 June 2007.
 Guang, L. and Assi, C. (2006) Mitigating Smart Selfish MAC Layer Misbehavior in Ad Hoc Networks. 2006 IEEE International Conference on Wireless and Mobile Computing, Networking and Communications IEEE, 19-21 June 2006.
 Akyildiz, I.F., et al. (2006) NeXt Generation/Dynamic Spectrum Access/Cognitive Radio Wireless Networks: A Survey. Computer Networks, 50, 2127-2159.
 Chen, R.L., Park, J.-M. and Reed, J.H. (2008) Defense against Primary User Emulation Attacks in Cognitive Radio Networks. IEEE Journal on Selected Areas in Communications, 26, 25-37.
 Wang, W.K., et al. (2009) Attack-Proof Collaborative Spectrum Sensing in Cognitive Radio Networks. 43rd Annual Conference on Information Sciences and Systems, 18-20 March 2009.
 Tan, X.B. and Zhang, H. (2012) A CORDIC-Jacobi Based Spectrum Sensing Algorithm for Cognitive Radio. KSII Transactions on Internet & Information Systems, 6, 1998-2016.
 Wagh, S., More, A. and Khavnekar, A. (2015) Identification of Selfish Attack in Cognitive Radio Ad-Hoc Networks. 2015 IEEE International Conference on Computational Intelligence and Computing Research (ICCIC).