International Journal of Conceptions on Computing and Information Technology Vol. 3, Issue. 1, April’ 2015; ISSN: 2345 - 9808 A Review on Intrusion Detection Systems for Securing MANETS V L Pavani Prof. B Satyanarayana Research Scholar, Dept.of Computer Science & Technology Sri Krishnadevaraya University Anantapur, Andhra Pradesh, India [email protected] Professor, Dept. of Computer Science & Technology Sri Krishnadevaraya University Anantapur, Andhra Pradesh, India Abstract— MANET is an infrastructure less network made up of self-configured nodes that can serve as communication media in case of emergencies. Due to their popularity MANETs are becoming ubiquitous. The nodes in MANET are resource constrained and vulnerable to various kinds of attacks such as gray hole, black hole, modification or Sybil attack, wormhole attack, and byzantine attack. Many intrusion detection systems came into existence to safeguard MANET from these attacks. The knowhow required to secure MANET communications can help organizations to be aware of security issues and can make well informed decisions to protect communications in MANET. Towards this end, in this paper, we focus on reviewing present state-of-the-art on intrusion detection systems for MANETs that cater to preventing various kinds of attacks launched by adversaries. Keywords- MANET, vulnerabilities, intrusion detection system, attacks I. INTRODUCTION Mantes are self managed networks. All the dynamic mobile nodes are connected randomly. Almost all the manets have dynamic nature , wireless technology , infrastructure less technology, Self configuration feature, open medium, distributed nature, changing topologies, resource constraints, lack of centralized administration, etc These features made manets vulnerable to different types of security threats . The Vulnerabilities may also be caused by the resource constraints of the participating nodes and nature of communication. The attacks may be passive or active attacks. Attacks may be eves dropping, inserting erroneous information regarding routing to create looping of routes, generating non functional or non existing routes ,gray hole, black hole, modification or Sybil attack , wormhole attack ,byzantine attack, etc. Attackers can also be classified as external or internal. External attackers modify the routes where as the internal attackers made the node as compromised nodes. So, there is a much need for most secure, powerful, efficient, reliable, adaptable, maintainable, portable , distributed, collaborative, independent and optimal Intrusion Detection System. In this paper we throw light into the present state-of-the-art on intrusion detection systems for MANETs and finding gaps in the research in safeguarding MANET communications. The section II of the paper focuses on various kinds of IDS that cater to handling different attacks on MANETs and section III contains the research gaps found . II. VARIOUS INTRUSION DETECTION SYSTEMS A. IDS for Sink hole, Worm hole and Block hole attacks Tseng and Culpepper (2005) [1] considered sequence number discontinuity and route add ratio as two indicators in a sinkhole Intrusion detection for manets using Dynamic Source Routing protocol. Sequence number discontinuity is detected by 3-tuple and route add ratio is computed by using route add counters of the nodes. Enhancing the above Kim, Han and Kim (2010) [2] proposed another sink hole detection algorithm. Here malicious node does not identify that the sink hole detection process is active. Sink hole indicator is used to indicate the sink hole. As soon as the sink hole indicator indicates a sink hole the algorithm is started automatically. Qian et al.(2007) [3] identified that multipath routing is vulnerable to worm hole attack and proposed a statistical analysis of multipath which is used to detect attacks and to identify malicious nodes. Tao Song (2007) [4]used System healthy intrusion monitoring, dynamic registration and configuration protocol to define the global security of a network to improve the healthy behavior of individual nodes. Intrusion Detection is done by monitoring the local nodes to achieve global integrity of network routing information. Stafrace (2010) [5] implemented a new mechanism for detection the sinkhole attacks by adopting the concepts of military structure and operation tactics and divided the algorithm into two phases. In first phase reconnaissance process is used a Command Post (CP) is set on all the nodes along the route from source to destination which consists intelligence process(INTEL) and next CP initiates phase where all the squad work together to detect sink hole nodes. Ming -Yang Su (2011) [6] talked about the block hole attacks and implemented a Anti Block hole Mechanism for detecting the suspicious nodes. Suspicious nodes are identified basing on the routing messages transmitted. If it exceeds some threshold vales that the node is suspected and that node details are broadcasted to all to nodes in order the add the node in the Block table. 83 | 9 0 International Journal of Conceptions on Computing and Information Technology Vol. 3, Issue. 1, April’ 2015; ISSN: 2345 - 9808 B. IDS based on Energy Constraints Kim, Kim and Kim (2006) [7] considered the Network Survivability is an important issue and suggested by choosing the monitor node based on the nodes battery power to enhance the network life time. Otrok et al. (2008) [8] implemented an IDS by electing a leader .Instead of selecting a leader randomly; a leader node is elected basing on certain properties such as node with most remaining energy to maintain the balance in resource consumption. Selfish nodes which provide wrong details such as fake value of energy remained, not running the IDS even though electing as a leader are also identified by using a catch and punish scheme by randomly selecting the checker nodes who monitors the actions of the leader. Any normal IDS runs without considering the energy constraint leading to a serious problem when a node looses all its energy for running IDS. To overcome this Cheng and Tseng (2011) [9], proposes a CAIDS model. It provides and intelligent mechanism for running the IDS. It not only deals with security threats but also the residual energy and traffic loading. Wang, Wang, Wang and Wang (2013)[10] focused mainly on wireless link performance centric scheme rather than security centric design approach and proposed a new technique called network tomography. This technique is designed by combining anomaly detection with inference techniques. A novel spatial time model to identify network topology and an energy aware algorithm to sponsor system services are proposed. Ramachandran el .at (2008) [11] used a task allocation algorithm where all the possible nodes are chosen basing the battery power levels and past history. C. Intrusion Detection System using layered architecture: Komninos et al.(2007)[31] focused on Security challenges in Intrusion Detection and authentication are identified . Two phase detection procedure is implemented. In phase one only the authenticated nodes are allowed to enter into Manet and in phase two compromised nodes are detected. Further this work is extended (2007) [32]. In phase one ,true communication nodes are identified by using the challenge response protocols based on symmetric key techniques and in phase two ,nodes are identified by using the challenge response protocols based on public key techniques .Cabrera el. at(2008)[13] introduced node- cluster- manager , a three level hierarchical system for distributed anomaly based IDS. For every‘t’ sec data related to network is acquired at each individual node by allowing the node to work normally. Komninis along with Douligeris extended his work on IDS and further proposed a LIDF (layered intrusion detection framework) which now consists of a three layers collection, detection and alert modules. This LIDF is implemented in OSI link layer and network layer operations. Ramachandran el .at (2008) [11] also used two layered two tier architecture. Main aim is to converge to an optimal solution. Stafrace(2010)[5] also used two phase mechanism for detecting sink hole attack. Shyu et al. (2007) [15] also designed the IDS by dividing it into the host and classification layers. D. Distributed Multi Agent based Intrusion Detection System: Shyu et al. (2007) [15] worked on distributed multi agent IDS architecture which consists of the two layers , the host layer and the classification layer. A set of host agents constitutes a host layer which collects the information regarding the network connections and divides them into normal or abnormal using ASEM anomaly detection scheme. In the classification layer multiple host agents are connected to a classification agents Again the host agents rely on their classification agents and these classification agents rely on manager agent which acts a central point. Thus ongoing threats are communicated in the same hierarchy in order to provide awareness in ongoing threat to all sub managers, classification agents and host agents. Kozushko (2003) [16] also implemented Distributed network based IDA. E. Securing the Manets by using both the IDS and Secure routing protocols: By Patwardhan et al. (2008) [17] MANET is secured by using both IDS and Secure Routing protocols. A node is not considered as a malicious node just by dropping packets .If the dropping of packets reaches its threshold value then that node is suspected. In case of routing protocols AODV protocol is considered. Generally the AODV provides route discovery and maintenance of local connectivity .A SecAODV protocol is implemented which provides a secure routing by binding IPV6 addresses and RSA keys. A node also monitors the neighboring nodes within the same radio range. F. IDS using game theory: Otrok et al. (2008) [8] used a co-operative game theoretic model is proposed to detect the false positive rate of the checkers. The efficiency of the IDS is also increased by formulating the zero sum non co-operative game and this game is solved by using Bayesian Nash Equilibrium. G. Trust Based IDS: Razak et al. (2008) [18] proposed a reliable intrusion mechanism that detects the attempts of attacks and at the same time reducing the false alarms raised. The IDS frame work implemented is based on two types of trust relationships, namely direct and indirect friendships. Global detection and response mechanism is also implemented by sharing the audit data sources with other nodes. Cho, Swami and Chen (2012)[19] considered the trust management with the concepts such as trust is dynamic not static , trust is subjective, trust is not necessarily a transitive, trust is asymmetry and the trust is context dependent. Initially in a group trust among the nodes is developed by using the historical information and authentication by challenge response process. From then onwards protocol generates the trust metrics of other nodes basing on social factors such as friendship, honesty, privacy, etc and QOS such as energy, computational power, radio range, etc. Further updates are also done continuously and a trust path is generated. Xia el, at (2013) [20] implemented a trust model for generating the optimal trust worthy routes in a single route 84 | 9 0 International Journal of Conceptions on Computing and Information Technology Vol. 3, Issue. 1, April’ 2015; ISSN: 2345 - 9808 discovery. Trusts are classified into historical, current and route trusts .Packets is divided into control and data packets. Forwarding ratio of two packets CFR (control packets forwarding ration) & DFR (data packets forwarding ratio) are calculated and the results are maintained in Trust record list at the each node. Node’s current trust is computed by using fuzzy logic rules prediction method by considering the historical values and also the current Values of the node such as battery power, local memory, DFR, CFR, band width. Etc. A source node initiates the route discovery process. If more than one route are discovered and all the routes meet the required trust level then the route with smallest hop count is selected. Chen, Guo, Bao and Cho (2014) [21] integrated Social trust and QoS trust and designed a protocol called SQTrust. The protocol was designed in such a way that the trust biased is minimized and the application performance is maximized. Among all, intimacy, honesty in social metrics where as competence and compliance in QoS metrics are considered. The trust level of the node is in between 0 and 1, indicating 1- complete trust, 0.5- ignorance and 0- distrust are computed by using the SPN techniques. H. Mathematical and statistical methods: Joseph el. at (2008) [22] discussed about the loop holes in IDS instead of discussing about different types of IDS in order to obtain the feasibility. The phenomenon called Base –rate fallacy means the efficiency of IDS also depends on probability of occurrence of malicious nodes and used Bayes theorem and conditional probability. In logical rule based technique theoretical limitations are considered .OSLR adhoc routing algorithm is used for comparing of different behavioral patterns is determined by using statistical methods. I. IDS using artificial intelligence techniques: Sen and Clark (2011) [23] used various artificial intelligence techniques like genetic programming or genetic algorithms and grammar evolution for implementing IDS in Manets. These are mainly used to detect adhoc flooding and route disruption attacks. This research area is loosely based on Darwinian Survival of the fittest and fitness of IDS is calculated. Shamshirband el. at (2013) [24] start working by reviewing all the previously works done and categorizing different Intrusion detection and prevention techniques. The main categories are traditional artificial intelligence, computational intelligence and multi agent based computational intelligence. By evaluating the performances and limitations, combining the features of computational intelligence and Multi agent based computational intelligence a new IDPS called Co-WIDPS (Cooperative Based intrusion detection and prevention system) architecture was designed. Komninis along with Douligeris (2009) [12] used Lagrange interpolation .The detection module checks whether the polynomial converges, and that node is considered as a compromised node. As soon as a new node enters it has to prove itself that it is not the compromised node if not the alert module generates an alarm. It uses linear threshold scheme. J. Work related to Comparison of different IDS: Kozushko (2003) [16]discussed host based and network based intrusion detection system by considering the life cycle of a network packet and divided the ID Architecture into Distributed network based IDA and centralized Host based . Pros and cons of both the architectures are compared. Ramachandran el .at (2008) [11] discussed about various types of IDS such as host based, network based, hierarchical , distributed co-operative based etc. but considered the only the agent based IDS. The IDS was designed by two tier architecture. Xenakis (2013) [25] compared different IDS‘s in stand alone, co-operative, multilayer co-operative, friend assisted, layered, Hierarchical, cluster based, and derived their strengths and weaknesses. Comparison of the deployment, architectural and operational characteristics, processing overhead, communication overhead, unfaired workload distribution, and other impacts on nodes. Comparison of detection of different types of attacks are also done and concluded by proposing the designs basing on the characteristics of manets. Pastrana el. at (2012) [26] extends his work for comparing different classification algorithms to detect malicious nodes in manets. Gaussian mixture model (GMM), Multilayer perception (MLP), linear model and support vector machines (SVM) and proved that SVM are MLP are the best. Again Behaviour of nodes using new classifiers based on genetic programming is compared with SVM and found that GP has low false alarm rate but detection rate is low. So, if an attack is found in advance such as Flooding GP algorithms work efficiently. K. IDS based on acknowledgement: According to Elhadi M .Shakshuki, Nat Kang, and Tarek R.Sheltami (2013) [29] MANETS dynamic and limited infrastructure leads to serious problems in critical situations. To increase the security in MANETS EAACK technique is implemented and three of the limitations of watch dog technique are resolved and forging of acknowledgement packet are also prevented. EAACK uses acknowledgements in the IDS. It consists of three major parts. First it uses end to end ACK scheme. If acknowledgement is not received then it detects that misbehaving nodes are present in the route and it switches from ACK to S-ACK (secure acknowledgement). In S-ACK malicious nodes are detected and misbehavior report is generated. To confirm this misbehavior report again it switches from S- ACK to MRA which detects the misbehaving nodes along with false misbehavior report. The idea of using Digital Signature for acknowledgement enhances the security. By extending the above work Basabba(2014) [27]implemented an ACA3K IDS . Similar to TWO ACK it works for three consecutive nodes and detects if any collaborative misbehaving nodes are present in the route path. 85 | 9 0 International Journal of Conceptions on Computing and Information Technology Vol. 3, Issue. 1, April’ 2015; ISSN: 2345 - 9808 III. RESEARCH GAPS FOUND The Qian [3], a multi-path routing protocol with intrusion detection, cannot run on other wireless protocols such AODV and AOMDV. However, SAM can be modified and applied to those protocols. SecAODV [28] consumes more energy resources for security routing of messages with larger header (control overhead). This needs to be addressed while increasing throughput and response times. Friend assisted IDS [12] can be improved further in order to explore the performance of IDS framework in the real world with respect to trust relationships. In the game-theoretic IDS [14] a node is considered either trusted or untrusted. However, it can be enhanced using a quantitative approach by giving rating to nodes that behave genuinely. The rating can provide more control and flexibility in IDS. FORM [15] can be enhanced further in order to improve prediction accuracy. In [16] it is explored that local knowledge is insufficient for accurate detection of intrusions. Therefore, it is needed to get enhanced by studying cross-layer approaches for continuous learning and adaptation to handle new attacks. Running HybrIDS [25] on distributed network test beds can improve its usefulness. EAACK [29] is a very effective IDS. However, it can be improved further in the areas like adapting hybrid cryptographic techniques, eliminating predistribution of keys, and testing EAACK in real world environments. SQTrust uses persistent attack models. However, it can be exposed to various other attack models such as fuzzy failure criteria, insidious attacks, opportunistic, and random models. Hassanzadeh et al. (2014) [30] proposed an IDS which was both traffic and resource aware. Experiments were made with Wireless Mesh Networks (WMNs). RAPID is another IDS explored in [68] for traffic aware intrusion detection. Basabba et al. (2014)[27]proposed an A3ACKs IDS which works when nodes are under different mobility. Assumptions: Even through powerful Intrusion detection systems are being implemented and used still cannot provide security to manets because of its openness features. The Intrusion Detection systems are communicating with the nodes in the radio range to detect the compromised or selfish nodes. Most of the acknowledgement based Intrusion Detection systems [29] are also communicating with the nodes by sending control packets to detect the non legitimate nodes[27]. Other Intrusion Detection systems are also communicating with the nodes in the radio range to detect the compromised or selfish nodes. It further increases the routing overhead. By considering all the above factors it is assumed that Trust Based Intrusion Systems are preferred. The trust values can be generated by using both direct and indirect measurements and updated by using moving average method. It is also predicted that we can decrease the routing overhead and computational complexity to some extent. In case of emergencies such as wars, disasters, etc and if most of the nodes are compromised where the secure communication plays a key role only the nodes with high trust values communication. can be used to provide secure TABLE.1: SUMMARY OF INTRUSION DETECTION SYSTEMS Author (s) Year Algorithm/Technique Tao Song [9] 2007 System Health and Intrusion Monitoring Razak et al. [13] 2008 ADCLI and ADCLU algorithms Otrok et al. [14] 2008 Game-theoretic IDS H.Chris Tseng and B. Jack Culpepper [28] Gisung Kim, Younggoo Han and Sehun Kim Protocol Dynamic Registration and Configuration Protocol Routing protocol independent solution Routing protocols Study Remarks Simulation and Empirical. Simulation Simulation 2005 Sequence Number discontinuity and Route add counters Dynamic Source Routing Protocol Simulation 2010 Sink hole detection algorithm Dynamic Routing Protocol Simulation Collaborative techniques are used in IDS Energy efficient leader election Two indicators of sink hole are proposed and analysed. Sinkhole detection time and rate are decreased. IV. CONCLUSION AND FUTURE WORK In this paper we studied MANETs and their security issues. It throws light into the present state-of-the-art of the intrusion detection systems that have been employed to safeguard MANETs in the real world. We strongly felt that there is room for further research as MANETs became popular and ubiquitous. Towards this end, this paper focuses on reviewing intrusion detection systems and finds the gaps in the research. As security can never be built, we strive to investigate the possibilities to improve the security of MANETs. This research can be further extended to investigate hybrid security mechanisms and more efficient trust based intrusion detection models for providing fool proof security to MANETs. REFERENCES [1] [2] [3] [4] [5] 86 | 9 0 H. Chris Tseng and B. Jack Culpepper. (2005). Sinkhole intrusion in mobile ad hoc networks: The problem and some detection indicators.Computers & Security. ELSEVIER 24 (n.d), 561-570. Gisung Kim∗,Younggoo Han,Sehun Kim.(2010). A cooperative sinkhole detection method for mobile adhoc networks. Int. J. Electron. Commun. ( AEU. 0 (0), p390-397. Lijun Qiana,_, Ning Songa, Xiangfang Lib (2007) , Detection of wormhole attacks in multi-path Routed wireless ad hoc networks: A statistical Analysis approach Multipath routing . Journal of Network And Computer Applications 30 (2007) 308–330. Tao Song. (2007). Formal Reasoning about Intrusion Detection Systems.Computer Science. . (n.d), p-1-206. Stefan K. Stafrace and Nick Antonopoulos. (2010). Military tactics in agent-based sinkhole attack detection for wireless ad hoc networks ,Computer Communications 33 (2010) 619–638 elsevier. 33 (n.d), p619–638. International Journal of Conceptions on Computing and Information Technology Vol. 3, Issue. 1, April’ 2015; ISSN: 2345 - 9808 [6] [7] [8] [9] [10] [11] [12] [13] [14] [15] [16] [17] [18] [19] Ming-Yang Su ⇑. (2011). Prevention of selective black hole attacks on mobile ad hoc networks through intrusion detection systems. Elsevier. Computer Communications 34 (2011) 107–117 Hyunwoo Kima,∗, Dongwoo Kimb, Sehun Kimc. (2006). Lifetimeenhancing selection of monitoring nodes for intrusion detection in mobile ad hoc networks. Elsevier. Int. J. Electron. Commun. (AEÜ) 60 (2006) p248-250. Hadi Otrok , Noman Mohammed, Lingyu Wang, Mourad Debbabi and Prabir Bhattacharya. (2008). A game-theoretic intrusion detection model for mobile ad hoc networks. Elsevier . Computer Communications– 31 (n.d), p-708–721. Bo-Chao Cheng a,*, Ryh-Yuh Tseng b. (2011). A Context Adaptive Intrusion Detection System for MANET. Elsevier . 0 (0), p310-318. Wei Wang a,b, Huiran Wanga, Beizhan Wangc,⇑, Yaping Wangd, Jiajun Wangc. (2013). Energy-aware and self-adaptive anomaly detection scheme based on network tomography in mobile ad hoc networks. Elsevier. 0 (0), p580-602. Chandrasekar Ramachandran , Sudip Misra and Mohammad S. Obaidat. (2008). FORK: A novel two-pronged strategy for an agent-based intrusion detection scheme in ad-hoc networks. elsevier. 31 (n.d), p3855–3869. Nikos Komninos a,*, Christos Douligeris. (2009). LIDF: Layered intrusion detection framework for ad-hoc networks. elsevier. 7 (n.d), p171-182 Joa B.D. Cabrera , Carlos Gutie´rrez, Raman K. Mehra. (2008). Ensemble methods for anomaly detection and distributed intrusion detection in Mobile Ad-Hoc Networks. elsevier. 9 (n.d), p-96–119. Hadi Otrok , Noman Mohammed, Lingyu Wang, Mourad Debbabi and Prabir Bhattacharya. (2008). A game-theoretic intrusion detection model for mobile ad hoc networks. elsevier. 31 (n.d), p-708–721. MEI-LING SHYU, THIAGO QUIRINO, and ZONGXING XIE. (2007). Network Intrusion Detection through Adaptive Sub-Eigenspace Modeling in Multiagent Systems. ACMTransactions on Autonomous and Adaptive Systems. 2 (n.d), p-1-37. Harley Kozushko. (2013). Intrusion Detection: Host-Based and Network-Based Intrusion Detection Systems. Independent Study. 11 (n.d), p-1-23 A. Patwardhan , J. Parker , M. Iorga , A. Joshi , T. Karygiannis and Y. Yesha. (2008). Threshold-based intrusion detection in ad hoc networks and secure AODV. elsevier. 6 (n.d), p-578–599. S.A. Razak , S.M. Furnell , N.L. Clarke and P.J. Brooke. (2008). Friendassisted intrusion detection and response mechanisms for mobile ad hoc networks. elsevier. 6 (n.d), p-1151–1167. Jin-Hee Cho a,n, Ananthram Swami a, Ing-Ray Chen b,1. (2012). Modeling and analysis of trust management with trust chain optimization in mobile ad hoc networks. Elsevier. 0 (0), p1001-1012. [20] Hui Xia , Zhiping Jia , Xin , Lei Ju , Edwin H.-M. Sha. (2013). Trust prediction and trust-based source routing in mobile ad hoc networks.elsevier. 11 (n.d), p-2096-2114. [21] Ing-Ray Chen a,⇑, Jia Guo a, Fenye Bao a, Jin-Hee Cho b. (2014). Trust management in mobile ad hoc networks for bias minimization and application performance maximization. Elsevier. 0 (0), p59-74. [22] John Felix Charles Joseph , Amitabha Das , Boon-Chong Seet and BuSung Lee . (2008). Opening the Pandora’s Box: Exploring the fundamental limitations of designing intrusion detection for MANET routing attacks. elsevier. 31 (n.d), p-3178–3189. [23] Sevil Sen and John A. Clark. (2011). Evolutionary computation techniques for intrusion detection in mobile ad hoc networks. elsevier. 55 (n.d), p-3441–3457 [24] Shahaboddin Shamshirband , NorBadrulAnuar , MissLaihaMatKiah and AhmedPatel. (2013). Anappraisalanddesignofamultiagentsystembasedcooperative wirelessintrusiondetectioncomputationalintelligencetechnique. elsevier. 26 (n.d), p-2105-2127 [25] Christos Xenakis ,Christoforos Panos and Ioannis Stavrakakis . (2013). A comparative evaluation of intrusion detection architectures for mobile ad hoc networks. elsevier. 30 (n.d), p-63-80 [26] Sergio Pastrana a,⇑, Aikaterini Mitrokotsa b, Agustin Orfila a, Pedro Peris-Lopez a (July 2012) Evaluation of classification algorithms for intrusion detection in MANETs. , Knowledge-Based Systems 36 (2012) 217–225 [27] Abdulsalam Basabaa, Tarek Sheltami, and Elhadi Shakshuki (2014). Implementation of A3ACKs Intrusion Detection System under various mobility speeds [28] H. Chris Tseng a,*, B. Jack Culpepper b. (2005). Sinkhole intrusion in mobile ad hoc networks: The problem and some detection indicators. Elsevier. 0 (0), p561-570 [29] Elhadi M. Shakshuki, Nan Kang, and Tarek R. Sheltami. (2013). EAACK—A Secure Intrusion-Detection System for MANETs. IEEE. 60 (3), p1089-1099. [30] Amin Hassanzadeh, Ala Altaweel, Radu Stoleru. (2014). Traffic-andresource-aware intrusion detection in wireless mesh networks. elsevier. 21 (.), p-18-41. [31] Nikos Komninos a,*, Dimitrios D. Vergadosa, Christos Douligeris. (2007 ) Authentication in a layered security approach for mobile ad hoc networks , computers & s e c u rity 2 6 ( 2 0 0 7 ) 3 7 3 – 3 8 0 [32] Nikos Komninos a,*, Christos Douligeris. (2007). Detecting unauthorized and compromised nodes in mobile ad hoc networks elsevier. Ad Hoc Networks 5 (2007) 289–2987 87 | 9 0
© Copyright 2025 Paperzz