Int. J. Intelligent Information and Database Systems, Vol. X, No. Y, xxxx
Biological inspired secure autonomous routing mechanism for wireless sensor networks Kashif Saleem*, Norsheila Fisal, Muhammad Sharil Abdullah and Sharifah Hafizah Syed Ariffin Universiti Teknologi Malaysia, 81300-Skudai, Johor, Malaysia E-mail:
[email protected] E-mail:
[email protected] E-mail:
[email protected] E-mail:
[email protected] *Corresponding author Abstract: The field of wireless sensor network (WSN) is an important and challenging research area today. Advancements in sensor networks enable a wide range of environmental monitoring applications. Multihop routing in WSN is affected by new nodes constantly entering/leaving. Moreover, secure routing is a difficult problem due to the resource limitations in WSN. Thus, biological inspired algorithms are reviewed and enhanced to tackle the problems. Ant routing and human security system have shown an excellent performance. Certain parameters as energy level, velocity, packet reception, dropping, mismatch rates and packet sending power are considered while making decision. The decision will come up with the optimal route and also to take best action against security attacks. In this paper, the design and initial work of BIOlogical Inspired Secure Autonomous Routing Protocol (BIOSARP) is presented. The proposed bio-inspired mechanism will meet the enhanced WSN requirements, including better delivery ratio, less energy consumption and routing overhead. Keywords: ant colony optimisation; ACO; human immune system; multi-agent system; optimal route; self-healing; self-organised; wireless sensor network; WSN; autonomous; biological; database systems; intelligent information. Reference to this paper should be made as follows: Saleem, K., Fisal, N., Abdullah, M.S. and Ariffin, S.H.S. (xxxx) ‘Biological inspired secure autonomous routing mechanism for wireless sensor networks’, Int. J. Intelligent Information and Database Systems, Vol. X, No. Y, pp.000–000. Biographical notes: Kashif Saleem received his BSc in Computer Science from Allama Iqbal Open University, Pakistan in 2002, his Post Graduate Diploma in Computer Technology and Communication from Government College University, Pakistan in 2004, and his Master of Engineering Degree in Electrical (Electronics and Telecommunication) from Universiti Teknologi Malaysia (UTM) in 2007. Currently, he is pursuing his PhD in Electrical Engineering at the Faculty of Electrical Engineering, Universiti Teknologi Malaysia (UTM), under the supervision of Prof. Dr. Norsheila Fisal. Norsheila Fisal received her BSc in Electronic Communication from University of Salford, Manchester, UK in 1984, and her MSc in Telecommunication Technology and her PhD in Data Communication from University of Aston, Copyright © 200x Inderscience Enterprises Ltd.
1
2
K. Saleem et al. Birmingham, UK in 1986 and 1993, respectively. Currently, she is a Professor in the Faculty of Electrical Engineering, University Technology Malaysia and the Director of Telematic Research Group. Mohd Sharil Abdullah received his Diploma in Electrical Engineering (Communications), degree in Electrical Engineering (Telecommunications) and Master in Electrical Engineering (Electronic and Telecommunications) from Universiti Teknologi Malaysia (UTM), in 2000, 2003 and 2009, respectively. Since 2007, he has been with Telematic Research Group at the Faculty of Electrical Engineering, University Technology Malaysia. His expertises are in wireless ad hoc network, wireless sensor network, data communications and information theory. His current research is in data compression using Slepian-Wolf theorem for wireless sensor network. Sharifah Hafizah Syed Ariffin received her BEng in Electronic and Communication Engineering from London Metropolitan University, London, UK and her MEE by research (mobility management in wireless telecommunication) from University Technology Malaysia (UTM), Malaysia, in 1997 and 2001, respectively. She obtained her PhD in Telecommunication (accelerated simulation) from Queen Mary University, London, UK, in 2006. She is a Senior Lecturer at UTM and her research interests include network modelling and performance, accelerated simulation, self similar traffic and priority scheduling.
1
Introduction
Wireless communication plays an important role these days in the sector of telecommunication (Chowdary, 2002) and has huge importance for future research. There has been an exponential growth in wireless communication due to the development of different devices and applications. In addition, there is an explosive increase in integration and convergence of different heterogonous wireless networks in order to ensure effective and efficient communication. These technologies primarily includes: wireless wide area networks (WWANs), wireless local area networks (WLANs), wireless personal area networks (WPANs), and the internet. The cellular networks can be classified under the WWAN, bluetooth and ultra wide bands (UWB) classified as WPANs, and finally the WLANs and high performance radio local area networks (HiperLANs) belongs to the WLAN class. Researchers at the University of California, Berkeley recently develop a new approach in wireless system design: one that involves low-cost embedded devices that can be implemented for a variety of applications (Cerpa et al., 2005). These small and low cost sensor nodes became technically and economically feasible (Al-Karak and Kamal, 2004). The sensor node is a miniaturised device which is equipped with sensors like temperature, humidity, light, sound, etc. Nevertheless, due to the extremely small architecture, the sensors lacks in storage space, energy supply and communication width. For example, a sensor typically has 8–120 KB of code memory and 512–4,096 bytes of data memory. The transmission bandwidth ranges from 10 kbps to 115 kbps. The sensor nodes are programmed to work in a self-organised way. Due to their autonomous capability, the sensor nodes can transfer the sensed data node by node to the destination known as sink node. Sink node is also called base station. The amount of base
Biological inspired secure autonomous routing mechanism for WSN
3
stations (like laptop, PDA, gateway to other networks, etc.) in the deployed network depends on the application requirements. Enormous numbers of these disposable sensor nodes come up with a wireless sensor network (WSN) as shown in Figure 1. Figure 1
Wireless sensor network
In sensing and monitoring systems new gadgets and software advancement are very frequently available to the end-user. This development subsequently increases the complexity of the network. Also some of infrastructure less WSNs deployment area is out of human reach. Since the nodes in a network can serve as routers and hosts, they can forward packets on behalf of the other nodes and run user applications (Frodigh et al., 2000). The resources in sensor node are so limited that every possible means of reducing the usage of these resources are aggressively required. In essence, sensor networks will provide the end user with intelligence and a better understanding of the environment. WSN demands self-organised communication which means the network can easily manage itself according to the changes in its environment. Furthermore, because of resource constraint and vulnerabilities of wireless communication, it is easier to suffer all kinds of attacks, if the sensor nodes are deployed in the unprotected/hostile environment (Ma et al., 2007). Some of these attacks include signal jamming, eavesdropping, tempering, spoofing, resource exhaustion, altered routing information, selective forwarding, sinkhole attacks, sybil attacks, wormhole attacks and flooding attacks (Karlof and Wagner, 2003). Since many sensor networks will be deployed in critical applications, security is essential. Unfortunately, security may be the most difficult problem to solve in WSN. Widespread acceptance and adoption of these protocols in real world WSNs would not be possible until their security aspects have thoroughly been investigated. However, security in these nature inspired routing protocols is still an open issue (Mazhar and Farooq, 2007). Most self-organised communication and coordination solutions do not address security, so it is easy for an adversary to exploit those implemented solutions on a given WSN. Basically, the research challenges for security in sensor networks are vast and difficult. Lightweight schemes are required. Solutions must exploit the nature of the sensor network, possibly related to issues such as most data is only valid for a short time (Felemban et al., 2005; Karlof and Wagner, 2003; Wood and Stankovic, 2002). Therefore, lightweight security is effective as individual nodes may possess little knowledge by themselves. Protecting the data aggregation function may also be possible. In summary, new ideas on the fundamental limits for security in these systems are
4
K. Saleem et al.
needed. The new mechanism can maintain the features of WSNs such as multihop routing and dynamically environmental changes in a complete autonomous mode. In order to address autonomous capability for multihop WSNs, it has been visualised that self-organised network application can understand the network operational objectives. Additionally, probabilistic methods that provide scalability and preventability can be found in nature and adapted to technology. Researchers anticipate self-organisation methods as the general solutions to the depicted communication issues in WSN and sensor and actor networks (SANETs). Centralised management and optimised control is replaced by methodologies that focus on local knowledge about the environment and adequate decision-making processes. Similar problems are known and well-studied in nature. Therefore, such biological solutions should be adapted to enhance the communication in ad hoc networks and WSN (Dressler, 2006). It is observed that various biological principles are capable to overcome the above adaptability issue. The area of bio-inspired network engineering which has the most well known approaches are swarm intelligence (ant colony, particle swarm), AIS and intercellular information exchange (molecular biology) (Balasubramaniam et al., 2006a 2006b; Boonma and Suzuki, 2008; Mazhar and Farooq, 2007). AIS have shown brilliant results for misbehaviour detection in WSNs. The principles of artificial immune system (AIS) help in designing and implementing the security framework. AIS learns the normal behaviour of the system and then monitors the system for occurrences of abnormal patterns. The most interesting working behaviour is the self-optimisation and learning process. Two immune responses were identified. The primary one is to launch a response to invading pathogens leading to an unspecific response (using leucoytes). In contrast, the secondary immune response remembers past encounters, i.e., it represents the immunologic memory. It allows a faster response the second time around showing a very specific response (using B-cells and T-cells). The system, therefore, has the ability to detect previously unknown attacks. Therefore, it seems obvious to apply the same mechanisms for self-organisation and self-healing operations in computer networks (Dressler, 2006). The author of (Mazhar and Farooq, 2007) proposed misbehaviour detection in nature inspired MANET protocol, BeeAdHoc. iNet proposed in Lee and Suzuki (2007) detects and eliminates the antigens (e.g., viruses) from the BiSNET/e enabled networks. This paper proposes a biological inspired self-organised secure autonomous routing protocol (BIOSARP) for WSNs. Ant colony optimisation (ACO) method is utilised for the optimum route discovery in multihop WSN. On network layer the optimal route is discovered through ACO as described in Saleem et al. (2009b). This routing algorithm is further enriched with a self-security mechanism inspired by AIS. The security mechanism is developed to secure the WSN from the most common network layer attacks such as sink hole attack, select forward, black hole, message alter attack, hello attack and worm hole. The preliminary report on this work may be seen in Saleem et al. (2009a). Security method is inspired by immune system (IS) to perform an autonomous protection over WSN. Fault and anomaly detection agent-based system (Ma et al., 2007) has been acquired from AIS. The above mentioned techniques are accomplished by assigning each procedure to the group of agents (Jiang et al., 2005). The agents such as search agent, data agent, etc. works in a decentralised way to collect data and/or detect an event on individual nodes. Once collected, it is transferred securely to the required destination/destinations through
Biological inspired secure autonomous routing mechanism for WSN
5
multihop communication. While moving from one node to another the agent checks characteristics of next node with the pre-initialised table. Over autonomous security is based on packet receiving rate, packet dropping rate, packet mismatch rate and packet sending power metrics. Relying on these parameters, certain alarm is generated by the agent or agents. The actions are applied according to the type of alarm triggered. Eventually, the autonomous security mechanism comes up with the architecture which prevents the WSN from the regular network layer attacks. The next section reviews the related research for optimum route discovery through ACO and security of network routing protocol using the AIS approach. Section 3 describes the way to implement this autonomously secure routing mechanism. Section 4 shows the simulation and analysis. The conclusion and future work are stated under Section 5.
2
Related research
2.1 Overview of ant routing in WSN Ant colony algorithms were first proposed by Dorigo et al. (Chen et al., 2006) as a multi-agent approach to difficult combinatorial optimisation problems like the travelling salesman problem (TSP) and the quadratic assignment problem (QAP), and later introduced the ACO meta-heuristic. ACO algorithms are a class of constructive meta-heuristic algorithms that mimic the cooperative behaviour of real ants to achieve complex computations and have been proven to be very efficient to many different discrete optimisation problems. Many theoretical analyses related to ACO show that this optimisation can converge to the global optima with non-zero probability in the solution space (Stuetzle and Dorigo, 2002) and their performance have greatly matched many well-studied stochastic optimisation algorithms, for example, genetic algorithm, pattern search, GPASP, and annealing simulations (Chen and Nasser, 2006). Das et al. have given an online ACO algorithm using AntNet techniques for MSDC (Singh et al., 2004) which has been formalised to be a typically minimum Steiner tree problem. They also have proposed an improved algorithm by adding another type of ants, random ants, just like the newspaper deliverer, whose main task is to dissipate information gathered at the nodes among other neighbouring nodes. Practically, simulation results also show that their algorithms are significantly better than address-centric routing. In these proposed algorithms the forward ants normally spend a long time. There is a bug of dead lock in their algorithms. In their improved algorithm, a large amount of random ants are needed. Aghaei et al. (2007) propose two adaptive routing algorithms based on ant colony algorithm, the AR algorithm and the IAR algorithm. To check the suitability of ADR algorithm in the case of sensor networks, they modified the ADR algorithm (removing the queue parameters) and used their reinforcement learning (RL) concept and named it the AR algorithm. The AR algorithm did not result in optimum solution. In IAR algorithm by adding a coefficient, the cost between the neighbour node and the destination node, they further improve the AR algorithm. Wen et al. (2008) propose E&D ANTS based on energy*delay metrics for routing operations. Their main goal is to maintain network lifetime in maximum and propagation delay in minimum by using a
6
K. Saleem et al.
novel variation of RL. E&D ANTS results was evaluated with AntNet and AntChain schemes. Comparison of the most recent ANT-based routing in WSN: SC and (Okdem and Karaboga, 2006) depends on the energy metric while FF based on delay. IA and IAR is the modification of ADR which used a delay parameter in the queues to estimate RL factor. In FP they combine the forward ant and data ant to enhance the success rate. E&D ANT based on energy*delay metrics for routing operations. In our proposed BIOSARP, the best values of velocity, PRR and remaining power mechanism (Ali et al., 2008) are used to select forwarding node because velocity alone does not provide the information about link quality. The best link quality usually provides low packet loss and energy efficient (Zhao and Govindan, 2003). Another novel feature of BIOSARP is it utilises the remaining power parameter to select the forwarding candidate node. The remaining power assists the source node or intermediate node to distribute the forwarding load to all available forwarding candidates and hence avoid the routing holes problem. BIOSARP is enabled with unicast forwarding scheme to route the data towards the best and nearest destination.
2.2 Security in WSNs The attractive features of WSNs involved many researchers to work on various issues. In WSN the routing strategies are getting more attention and the other hand security issues are not taken under consideration up to the required level. It is imperative that the security concerns be addressed from the beginning of the system design (Walters and Liang, 2007). Many WSNs routing protocols are quite simple, and for this reason are sometimes even more susceptible to attacks against general ad-hoc routing protocols. In WSN, an adversary can either deploy his own node or compromise some nodes. The compromised node can take many actions to create a network layer attacks. Most network layer attacks against sensor networks fall into one of the following categories; spoofed, altered, or replayed routing information; selective forwarding; sinkhole attacks; sybil attacks; wormholes; hello flood attacks and acknowledgement spoofing (Karlof and Wagner, 2003). In the descriptions below, attacks which are based on manipulated sensor data are divided into two classes: that include attacks that try to manipulate user data directly and attacks that try to influence the underlying routing topology. Because of resource constraint and vulnerabilities of wireless communication, it is easier to suffer all kinds of attacks, as sensor nodes are mostly deployed in the unprotected/hostile environment. These attacks involve signal jamming and eavesdropping, tempering, spoofing, resource exhaustion, altered or replayed routing information, selective forwarding, sinkhole attacks, sybil attacks, wormhole attacks, flooding attacks and so on (Pathan et al., 2006). Many papers have proposed prevention countermeasures of these attacks and the majority of them are based on encryption and authentication. However, these prevention measures in WSN can reduce intrusion to some extent. In this case, intrusion detection system (IDS) can work as second secure defence of WSN to further reduce attacks and insulate attackers. In traditional networks, traffic and computation are typically monitored and analysed for anomalies at various concentration points. Though, this is often expensive in terms of network’s memory and energy consumption, as well as its inherently limited bandwidth. WSNs require a solution that is distributed and inexpensive in terms of communication,
Biological inspired secure autonomous routing mechanism for WSN
7
energy, and memory requirements. Therefore, these traditional techniques of IDS must be modified or new techniques must be developed to make intrusion detection work effectively in WSN.
2.2.1 Overview of IDS-based security IDS in traditional network have been widely proposed and applied. Siraj et al. (2004) presented a model of decision engine for intelligent IDS. This decision engine use fuzzy cognitive maps and fuzzy rule-bases for causal knowledge acquisition and to support the causal knowledge reasoning process. Harmer et al. (2002) presented artificial IS architecture for computer security. Albers and Camp (2002) proposed a kind of general intrusion detection architecture based on the implementation of a local intrusion detection system (LIDS) at each node. Kachirski and Guha (2003) proposed a multi-sensor IDS based on mobile agent technology. They divided the mobile agents into three kinds of agents: monitoring agent, decision agent and action agent. In WSN, Alpcan and Agah et al. proposed to adopt game theory for decision and analysis in intrusion detection of WSN (Alpcan and Basar, 2003; Das and Basu, 2004). They investigate the basic decision and analysis processes involved in information security and intrusion detection, as well as possible usage of game theory for developing a formal decision and control framework. Generic model for distributed IDSs was introduced by defining a network of sensors, and propose two simple, flexible, and easy-to-implement schemes utilising both cooperative and non-cooperative game theoretic concepts. ACO is also utilised for intrusion detection in Banerjee et al. (2005b, 2005a). The authors have introduced the concept of Tabu list, where for every session the list would like to store the pheromone trace or path that is prone to attack. In Gao et al. (2005), the ACO-based intrusion feature selection algorithm is proposed. The FDR is taken in as the heuristic information for ACO. The authors have adopted the least square-based SVM estimation to avoid training of a large number of SVM classifier. The results have been demonstrated, by which they have show the detected attacks as probe, DoS and U2R&R2L intrusions.
2.2.2 Overview of AIS-based security AIS are adaptive systems, inspired by theoretical immunology and observed immune functions, principles and models, which are applied to problem solving. Adaptability in the IS ensues from features such as learning and memory that endow the IS with the ability to fight a large variety of invaders (de Lemos et al., 2007). The application of AIS to fault tolerance was initially motivated by Avizienis (1997), who described the analogy between the IS and fault tolerance. Since then, several approaches have been proposed in literature that have applied AIS to problems related to both software and hardware fault tolerance (de Lemos et al., 2007). In Ma et al. (2007), the authors have proposed a novel IDS (SAID) to be suitable for deploying in WSN. SAID with three-logic-layer architecture adopt the merits of LIDS and distributive and cooperative IDS and is self-adaptive for intrusion detection of resource-constraint WSN. SAID can actively trigger agent evolution to more effectively prevent intrusion when WSN suffers unknown attacks. For distributive cooperation
8
K. Saleem et al.
attacks, these distributive mobile monitor agents will cooperatively collect abnormal information of network to help a correct intrusion decision. Knowledge base is deployed base station where the complex algorithm (e.g., genetic algorithm) for agent evolution can be computed and intrusion rules can be stored. de Lemos et al. (2007) detail the investigations undertaken to develop an immune-inspired adaptable error detection (AED) technique for ATMs. The proposed framework for AED consists of two levels of error detection. One level of the framework is local to a single ATM, while the other is network-wide adaptable error detection. In the given architecture, each ATM hosts a local AED, while the network-wide AED is implemented within the central management system. The implementation undertaken in this work was limited to the local AED. An AIS algorithm was found to possess these characteristics, and was evaluated by using relevant criteria that include: 1
classification performance of the algorithm in discriminating normal behaviours from potential failure behaviours
2
the measurement of the time interval between detection and the actual system failure.
Exposed results demonstrate that the described AED technique could detect an incipient system failure approximately 12 hours for one dataset, and two hours for a second dataset. Li et al. (2008) proposed a new group-based intrusion detection scheme which is a detection-based technique. In this scheme, the authors partition the sensor nodes in a network into a number of groups. The nodes in a group have the same sensing capability and are physically close to each other. And the proposed intrusion detection algorithm is scheduled to run for each group. Through experiments in which the authors use data released from the Intel Berkeley Research Lab, shown that their scheme can achieve a lower false alarm rate and a higher detection accuracy rate than the present intrusion detection schemes and would consume less power. In Prattipati and Hart (2008), they have reevaluates the AISEC with the different set of test e-mails and have also extended the method of Secker et al. (2003). While extension the authors have specifically addressed the certain objectives. Initially they investigate the sensitivity of the algorithms parameters to different datasets. Afterwards, test the ability of the algorithm to adapt to changing interests, by setting up a number of test scenarios in which the users interest in e-mails from a particular source changes from interesting to un-interesting (and vice versa) over a period of time. Subsequently, AISEC has been modified, while improving the speed of algorithm at which it adapts and the overall accuracy of the classification algorithm. In addition, they changed the AISEC in .NET as an add-in for Microsoft Outlook and named as AISEC-Outlook. When the user supplies negative feedback from misclassification of an item in the inbox, the e-mail is now added to the repository of B-Cells responsible for classifying mails. Based on this feedback, AISEC-Outlook rewards the B-Cells. Lee and Suzuki (2010) propose and evaluate a decentralised self-healing mechanism that detects and recovers from wormhole attacks. Upon detecting a wormhole attack, the proposed mechanism, called SWATa, isolates wormhole nodes from the network by eliminating links connected to them, and recovers the routing structure distorted by the wormhole nodes. SWAT is designed as a decentralised in network detection mechanism that uses network connectivity information only. Finally, they have implemented SWAT on MICA2 and TelosB motes and acquire the result under Tinyviz. The simulation results
Biological inspired secure autonomous routing mechanism for WSN
9
shown in this paper yields 100% wormhole attack detection, 0% false detection, 100% wormhole node isolation and 0% false isolation in dense networks. Through SWAT, they have come to tackle only wormhole attacks on WSN. Since the characteristics of WSN (e.g., resource constrains of sensor nodes, ad hoc mechanism, the sensor node may be static after deployment, etc.), most IDS for internet network or mobile ad hoc network cannot be applied in WSN well. Therefore, the researches in IDS of WSN are still at the beginning (Ma et al., 2007). The operation of BeeHive algorithm requires an initialisation phase (30 seconds) even before the AIS learning could start. It is followed by the learning (50 seconds) and protection phases to respectively learn the BeeHive normal behaviour and detect the routing attacks (Mazhar and Farooq, 2007). The existing security methods are not applicable for self-organise routing in WSNs because the execution time for one-hop is high and WSNs have density deployment where hundreds of nodes need more time to process security mechanism.
2.2.3 Overview of keying-based security Su et al. (2005) proposed two approaches to improve the security of clustering-based sensor networks: authentication-based intrusion prevention and energy-saving intrusion. The proposed authentication-based intrusion prevention is enhanced from μTESLA which uses one-time key chains. Therefore, each CH needs to be loosely time synchronised with its member nodes. All sensor nodes are loosely time synchronised. The synchronisation in WSNs is very hard and mostly because of huge number of nodes it is impossible to have an accurate synchronisation. Due to the factor of initialisation phase, WSN need security mechanism to be in operation before the network deployment. As stated in Lim (2008), node cloning attacks can be mounted only during deployment since a cloned node cannot initiate the protocol with success; it can be successfully connected only by acting as a responding node. Recent progress in implementation of elliptic curve cryptography (ECC) on sensors proves public key cryptography (PKC) is now feasible for resource constrained sensors (Wang et al., 2008). Given the efficient low-layer primitive in place, the high-layer PKC-based security scheme design in sensor networks, however, is not straightforward due to the special hardware characteristics and requirements of sensor networks. Therefore, the performance of PKC-based security schemes is still not well investigated. In Xing and Liu (2008), the scheme has been proposed which explores the superimposed s-disjunct code for a timely clone attack detection. A fingerprint can be easily encoded with a very short bit stream, which results in small message overhead. Their mechanism can identify cloned sensors with high detection accuracy at the expense of a very low communication/computation/storage overhead. Given scheme conducts fingerprint verification locally (via neighbouring nodes) and globally (via the base station) for each message broadcasted by any node, therefore, clone attackers can be detected in real-time. In Ali and Fisal (2008b), the authors present the security enhancement which uses the encryption and decryption with authentication of the packet header to supplement secure packet transfer. SRTLD solves the problem of producing real random number problem using random generator function encrypted with mathematical function. The output of random function is used to encrypt specific header fields in the packet such as source, destination addresses and packet ID. Moreover, the data authenticity problem is solved in SRTLD using authentication procedure applied after decryption. In this mechanism they
10
K. Saleem et al.
assume that each sensor node is static, aware of its location and the sink is a trusted computing base. The mechanism pairwise key establishment (PKE) based on transitory master keys as discussed in Lim (2008) is particularly useful for the purpose. LEAP++ consists of system setup, pairwise key establishment and authentication key disclosure. Security in natural inspired routing protocols is still an open issue (Mazhar and Farooq, 2007). Widespread acceptance and adoption of these protocols in real world wireless networks would not be possible until their security aspects have thoroughly been investigated. Comparison of the most common secure routing protocols in WSN: The MAC and physical layers are based on IEEE 802.15.4 which is designed for low rate communication such as WSN. Literature review concludes that the data security and routing design in WSN are not easy works due to the numerous constrains in WSN such as memory storage, power limitation and unreliable wireless communication. The aforementioned limitations should be considered when security based on routing is designed. Table 1
Comparison of the most recent security-based protocols
Title of the protocol
Tackled attacks
Su et al. (2005) by SU
Bogus routing information Hello floods
Limitations Sensor nodes cannot move and new sensor nodes cannot be added after the pairwise keys are established
Black hole Select forward BeeAIS SAID
Non-self antigens Worm hole
Assuming no attack in first 30 seconds Assumption of no attack while initialisation
Sybil Li et al. (2008) by LI
Fabric information Group-based intrusion mechanism and only internal detection attack Select forward Sink hole Hello attack Worm hole attack
Xing and Liu (2008)
Clone attacks
Have less overheads but tackling only kind
LEAP++
Clone attacks
Assuming NCC is willing invest large enough time.
DoS attacks Assume all nodes deployed at a time are Worm hole attacks programmed to start neighbour discovery at some delay time after being airdropped. SRTLD
Hello flood
Each sensor node is static and aware of its location.
Selective forwarding
Sink is a trusted computing base.
The security measures as we have discussed above have still lot of holes and could not tackle the most common WSN routing attacks. In front of the currently running security protocols, BIOSARP will protect WSN from the most common attacks yet stated (Karlof
Biological inspired secure autonomous routing mechanism for WSN
11
and Wagner, 2003; Du et al., 2006). Our autonomously secure routing mechanism BIOSARP will make security decisions based on packet reception rate, packet dropping rate, packet mismatch rate and packet sending power metrics. As further the security enhancement based on human nerve barrier system will also helps to cut down the feedbacks. This barrier system will help to differentiate between a good and malicious node to be or not to be a member of the running sensor network. Other security attacks and intrusions are handled by the artificial immune misbehaviour/IDS. Finally, BIOSARP routing scheme has three type of security: built-in security due to random selection for next hop, authentication-based security and intrusion detection. Hence, it is more difficult for an adversary to attack and intercept a message. Table 1 summarises the most common secure routing protocol in WSN.
3
Methodology
3.1 System design System design deals mainly with the development of state machine diagrams the routing management, neighbourhood management, energy management and security management as shown in Figure 2. Figure 2
System diagram
12
K. Saleem et al.
Routing management will be dependent mostly on forwarding metrics calculation. While establishing the forwarding procedure, the routing management will look for the next best node towards the destination through the routing table, available at every node. By acquiring the optimal route from the routing table, routing management will finalise the forwarding process. Otherwise, if it could not find next node towards destination, then routing management will call the process of neighbour discovery under neighbourhood management as shown in Figure 2. The neighbourhood management will then search for the best neighbouring node by calling method. Calling method take place through broadcasting hello massages. The node which broadcast will receive replies from the neighbouring nodes along with their characteristics. On the base of these replays it provides the final solution back to the routing management state. According to this solution the routing management state will update the routing table on current node. The key role of power management state is to check the remaining power and inform accordingly to the higher state. The power management state also can adjust the power for the transceiver according to the environmental conditions. Under this state the energy parameter is imported from the physical layer into the network layer. In wireless sensor node there are five levels of power transmission. At the time of forwarding the first level is utilised, but if node is out of reach then the power level is increased in stages. Helping neighbourhood management state in the energy aware route discovery and power level management is controlled by the power management state. Figure 3
Routing management
Biological inspired secure autonomous routing mechanism for WSN
13
Inside routing management the forwarding metrics calculation takes place as shown is Figure 3. The forwarding metrics as given in Table 2 are calculated to get the optimal route decision towards the destination. If the error occurred while processing this state, it will be control by routing problem handler as elaborated in Figure 3. The error can be like required neighbour not present or the best neighbouring node is lost or the required parameter is not there. Otherwise, if there is no error while forwarding calculation then the any-cast state will be called to forward the required packets. Table 2
Node 1 Node 2 . . . . Node n Figure 4
Routing metrics Velocity (End2End delay)
Energy
Link quality (PRR)
τ1 τ2 . . . . τn
η1 η2 . . . . ηn
ω1 ω2 . . . . ωn
Neighbour management
Common functions under neighbour management state are neighbour table maintenance, neighbour discovery, insert new neighbour, neighbour replacement, etc. as exposed in Figure 4. Main and most important thing the routing table is maintained via this state. If the best node towards the destination could not be found, the child state neighbour discovery is initiated. The explored new nodes will be checked with the old records by neighbour replacement process. While inserting a new record, the routing table space is first checked by neighbour table space state according to the wireless sensor node memory. Finally, the new record is inserted in routing table through insert new neighbour state.
14
K. Saleem et al.
While performing routing management the forward agents and search agents are also checking the characteristics of next node through the parameters as given in Table 3. With this check the abnormality over the network is detected. The alarm based on the type of abnormality is then generated by particular agents. The alarm message is then handled by the security management module. Under security module decision and defence agents are working. Decision agent takes the alarm message and makes the decision accordingly through equation (6). Security parameters
Table 3
Packet receiving rate
Energy
Packet mismatch rate
Packet receiving rate
Sink hole attack {Li, 2008 #116}
To make relationship with
Message alter attack {Li, 2008, #116}
To make relationship with
Node 1
x11
y11
x12
y12
Node 2
x12
y12
x22
y22
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
Node n
x1n
y1n
x2n
y2n
Tackled attacks
According to the information from the decision agents, the defence agents can visit the neighbour nodes of the attack node and take appropriate actions. These actions will include 1
requiring the neighbour nodes to low the priority or refuse relaying the packages from these attack nodes
2
telling the sending node another routing path in order to circumvent the attacker
3
repairing the attacked node by renewing the encryption keys of nodes.
By these ways, they can isolate the attack nodes successfully. The optimal forwarding is determined by ACO mechanism. ACO makes the optimal route decision based on the pheromone value acquired by probabilistic rule. The pheromone value is based on three metrics that are, velocity, PRR and remaining power mechanism as given in Table 2. The link quality of the wireless medium determines the performance of WSN. In design of BIOSARP, the link quality is considered in order to improve the delivery ratio and energy efficiency. It should be noted that the link quality is measured based on PRR to reflect the diverse link qualities within the transmission range. We have added second heuristic value ωij in probabilistic rule to determine the link quality of neighbouring nodes while making decision. The probabilistic rule is expressed mathematically as equation (1). α
pijk (t )
=
∑
β
ϑ
⎡⎣τ ij (t ) ⎤⎦ . ⎡⎣ηij (t ) ⎤⎦ . ⎡⎣ωij (t ) ⎤⎦ α
h∈ jik
β
ϑ
⎡⎣τ ij (t ) ⎤⎦ . ⎡⎣ηij (t ) ⎤⎦ . ⎡⎣ωij (t ) ⎤⎦
(1)
Biological inspired secure autonomous routing mechanism for WSN
15
pijk (t ) overall desirability for ant k located in city i to choose to move to city j
τij depends on the delay parameter ηij is a heuristic evaluation of edge (i,j) ωij is the 2nd heuristic evaluation of edge (i,j). α, β and ϑ are three parameters that control the relative weight of pheromone trail and heuristic values. Eventually, the routing management will forward a data packet to the neighbour that has an optimal pheromone value pijk (t ).
V/Vm calculate the value of τij, Vbatt/Vmbatt calculates the value of ηij and value of ωij is obtained by PRR. Where Vm is the maximum velocity of the RF signal that is equal to the speed of light, Vmbatt is the maximum battery voltage for sensor nodes and is equal to 3.6 volts (Ali et al., 2009). The determination of PRR, battery voltage (Vbatt) and packet velocity (V) is elaborated in the following sections. Velocity factor has been integrated to support real time traffic over WSN. The velocity factor depends on end to end delay from source to destination node. The maximum packet velocity (V) between a pair of nodes is calculated by the help of equation (2) (Ali and Fisal, 2008a). V=
d (S , N ) Delay ( S , N )
(2)
where d(S,N) is the one-hop distance between source node S and destination node N. The link quality of the wireless medium determines the performance of WSN. In designing BIOSARP, the link quality is considered in order to improve the delivery ratio and energy efficiency. It should be noted that the link quality is measured based on PRR to reflect the diverse link qualities within the transmission range. PRR is determined by equation (3) (Ali et al., 2008). ⎡ ⎛ 8 ⎞⎛ 1 ⎞ PRR = ⎢1 − ⎜ ⎟ ⎜ ⎟ ⎢⎣ ⎝ 15 ⎠ ⎝ 16 ⎠
⎛ ⎛ 1 ⎞ ⎞⎤ ⎛ 16 ⎞ (−1) ⎜ ⎟ exp ⎜ 20 SNR ⎜ − 1⎟ ⎟ ⎥ ⎝j ⎠ ⎝ j ⎠ ⎠ ⎥⎦ ⎝ j =2 16
∑
m
j
(3)
SNR is calculated in equation (4) (Ali et al., 2008). SNR = Pt − PL(d ) − Sr
(4)
where Pt is the transmitted power in dBm and Sr is the receiver’s sensitivity in dBm.
3.2 Security management ACO will be further enhancing with an additional security management module as shown in Figure 2. The security management module is based on AIS to self-secure the WSN from the foreign bodies or attacks. The AIS mechanisms discussed under related research section are very complex and their algorithms are huge for WSN. These mechanisms also need enormous database to maintain knowledge. Because of their bulky behaviour, we have divided the jobs over agents. Our agent-based autonomously secure mechanism works on two-layered architecture, agent layer on the top and WSN on bottom as shown
16
K. Saleem et al.
in Figure 5. The agent layer contains three types of agents, which are monitoring, decision and defence. As described in Figure 6 through state diagrams. Figure 5
Two layer security architecture for intrusion detection 1.1. Monitoring (forwarding and search ant) ant) Monitoring (forwarding and search
Defense 2.2. Decision Decision Abnormal 3.3. Defence information
Perform action Wireless sensor node
Figure 6
Decision and defence agent
Biological inspired secure autonomous routing mechanism for WSN
17
We have acquired statistical mechanism in the correlation coefficient as in Ma et al. (2007), to check the abnormality over the network. Before generating self-optimise routing decision, BIOSARP checks the abnormality in that region as shown in Figure 7. The error rate can be set or change according to the scenario. If application needs more accuracy, we raise the error rate value. When the algorithm found abnormality, it will goes for more thorough checking. While performing thorough checking, the algorithm will explore the neighbouring table. As, AIS explore the neighbouring table, it matches the neighbouring nodes values with the given threshold values. The threshold values are changed according to the characteristics of the sensor nodes. Figure 7
AIS algorithm
Check neighbour table If (no neighbour) then number of entries n = 0 Else number of entries n + 1 If (number of entries n is not equal to 0) Then Number of Entries = n Neighbouring Nodes Packet Reception Rate = x(n) Neighbouring Nodes Energy = y(n) Call Correlation function and check relation between x(n) and y(n) If (Relation is less then given rate or bigger then -ve error rate) Then (Abnormality found and checks neighbouring nodes) If (node values greater then and less then given threshold) then remove the current node Else (continue the data forwarding process) Else (continue the data forwarding process)
Monitor agent is functioning on every node over WSN. At the time of optimal selection process the monitor agent will detect the malicious behaviour of neighbour nodes. Behaviour is detected by making a relationship in between the parameters. The parameters on which the relationship is verified are packet receiving rate and packet mismatch rate as shown in Table 3. To simplify the pattern matching, the statistical matching rule is adopted (Ma et al., 2007). The correlation coefficient produces a number between –1 and 1 that relates how similar the two input sequences are. It is defined in equation (5) (Ma et al., 2007). N
X , Y = {0...255} , N = l / 8, ρ =
∑ (X n i =1
i
− X )(Yi − Y )
∑ ( X i − X ) ∑ in=1 (Yi − Y ) n i =1
2
2
(5)
If the relation is having more difference then the given percentage or value, the communication is blocked. At this situation, monitor agent call the decision agent. The decision agents are similar with B-Cell in IS. All kind of decision agents is distributive, mobile, cooperative and redundant. Thus, these decision agents can cooperate effectively
18
K. Saleem et al.
to make a correct decision for distribute attacks. The major objective of decision agents is to detect the existence of non-self patterns within a potentially large set of existing non-self patterns. The matching criteria depend on the equation (6) (Ma et al., 2007). ⎧malicous, f ( I ,α ) ≥ 1 − ε match( f , ε , I , D) = ⎨ ⎩benign, otherwise
(6)
where I = input string D = decision agent’s matching string f = matching function ε = matching threshold. If decision agent finds non-self node/nodes then it call defence agent to take certain actions against classified non-self nodes. The defence agents act somewhat similar as the antibody that is secreted by lymphocyte. Their function modules involve the self-copy, isolation and suicide. As soon as the security process is completed, the WSN communication is normalised. In addition, the network has been secured by random encryption system as explained in Ali and Fisal (2008b). This is inspired by human immune barrier system. In this algorithm, the security management also handles the encryption system process. The data and control packets are sent to security management module to execute the proposed encryption before the packet is unicast or broadcast. On the other hand, the packet received is sent to the security management module to perform the proposed decryption and authentication before the packet is processed by other modules. The encryption-based security in is based on work done on SRTLD (Ali and Fisal, 2008b). Since sensor nodes function as routers, the encryption and decryption with authentication process should be made at every hop for every forwarding packet in WSN. In output, BIOSARP encrypt specific header fields in the packet such as source, destination addresses and packet ID.
4
Simulation and analysis
4.1 Network topology To evaluate the above analysis, we use Network Simulator 2 (NS2) to construct the network topology as given in Figure 8. The topology is described as 121 nodes are distributed in an 80 m × 80 m region. Nodes numbered as 120, 110, 100 and 90 are the source nodes, node 0 is the sink and nodes numbered as 24, 25, 31 and 36 are adversary nodes. In this simulations study, the packet rate is ten packets per second while the end-to-end deadline and simulation time are 250 ms and 100 s respectively. CBR traffic is assumed in this simulation. The programme is written in NesC programming language to implement the biological inspired routing algorithm.
Biological inspired secure autonomous routing mechanism for WSN Figure 8
19
Network simulation grid with attackers (see online version for colours)
Notes: sink node (black), attackers (red), source node (blue) and sensor node (green).
In Figure 8, each link is bidirectional and the weighting value of the link depends on the power consumption (nJ/bit) and ant’s moving time delay (ms). The experimental parameters used to configure the system according to WSN are listed in Table 4. Table 4
System properties
Propagation model
Shadowing
Path loss exponent
2.45
Shadowing deviation (dB)
4.0
Reference distance (m)
1.0
Parameter
IEEE 802.15.4
phyType
Phy/WirelessPhy/802_15_4
macType Operation mode Ack
Mac/802_15_4 Non-Beacon (unslotted) Yes
CSThresh_
1.10765e-11
RXThresh_
1.10765e-11
freq_ Initial energy
2.4e+9 3.3 joule
Power transmission
1 mW
Transport layer
UDP
Traffic
CBR
20
K. Saleem et al.
Figure 9 shows the consequence of selective forwarding attack in WSN. Mark 1 in Figure 9 shows that the AIS security algorithm detects abnormality. When AIS found abnormal behaviour, the algorithm will then explore the neighbour table. While exploring AIS match the characteristics of neighbouring nodes with the given threshold. If any of the entry mismatches or crosses the given threshold, the entry is deleted from the neighbouring table. Like node 29 is removed as shown in Figure 9. The behaviour is same as the human body, like if IS finds any abnormal behaviour it reacts. In reaction, the IS removes cells, parts or even a complete organ. Figure 9
NS-2 showing the abnormality and actions taken against certain nodes (see online version for colours)
In Figure 9, mark 2 shows an unsecure packet received from the node 31. The encryption and decryption mechanism helps even at the beginning stage of network. As in the beginning stage AIS does not come up with its knowledge. Whenever an unsecure packet is received from the neighbouring nodes, without opening and reading it is dropped directly.
4.2 Simulation analysis The proposed BIOSARP routing protocol is studied through simulation process. Its performance is analysed and compared with SRTLD protocol, which is also based on link quality, velocity and energy. The simulation is also designed to evaluate the performance of the forwarding policies running in conjunction with various management policies. The simulation results in Figure 10a show that the delivery ratio of BIOSARP is higher by 50% compared to the SRTLD routing protocols. Table 5 also shows WSN lifetime is prolonged by 48% in BIOSARP compared to the SRTLD.
Biological inspired secure autonomous routing mechanism for WSN
21
BIOSARP consumes up to 50% less power compared to SRTLD protocols as shown in Figure 10b. The reduced power consumption is the consequence of sending and distributing the load throughout the neighbouring nodes. Table 5
Comparison WSN lifetime between BIOSARP and SRTLD
BIOSARP
SRTLD
Lifetime
439
224.5
Normalised lifetime
99%
51.02%
Figure 10 Comparison between BIOSARP and SRTLD routing at fixed packet rate: (a) delivery ratio (b) energy consumption
(a)
(b)
22
5
K. Saleem et al.
Conclusions and future work
Here, we have proposed a biological inspired secure autonomous routing mechanism named as BIOSARP for WSNs. The routing decision is based on ACO and the self-security on AIS. This decision depends on our use metrics. The proposed mechanism will successfully detect the non-self antigens and protect WSN against wormhole, sinkhole, sybil, selective forwarding and hello flood attacks. BIOSARP routing protocol enhances the previous works by Ali et al. (2009, 2008), Ali and Fisal (2008b), Ramadan (2009), Okdem and Karaboga (2009), and Wen et al. (2008) in order to achieve high delivery ratio and efficient power consumption. In general, the finding concludes that BIOSARP provides high delivery ratio with less power consumption as compared to the SRTLD routing protocol. Our proposal clearly demonstrates that AIS-based security has the potential to offer significantly higher performance in WSN due to its significantly less control, energy and computational cost. The efficient utilisation of these resources is a key challenge in WSNs. We also improve this AIS principle by adopting a special feature from human nerve structure called as barrier system. This would be a cardinal step in deploying BIOSARP in real world WSNs. Our immediate future work will involve building and testing the architecture by the direct implementation of ACS and AIS in real test bed experiment. Onwards, other ant colony variants such as negative reinforcement, max-min ant system will also be considered. The integrated algorithm of AIS will be also enhanced by adding more features by nature.
Acknowledgements The authors would like to thank the Ministry of Higher Education Malaysia for their full support and Research Management Center, Universiti Teknologi Malaysia (UTM) for their partial and the researchers in Telematic Research Group, UTM.
References Aghaei, R.G., Rahman, M.A., Gueaieb, W. and Saddik, A.E. (2007) ‘Ant colony-based reinforcement learning algorithm for routing in wireless sensor networks’, Instrumentation and Measurement Technology Conference, IEEE, IMTC Warsaw, Poland. Albers, P. and Camp, O. (2002) ‘Security in ad hoc networks: a general intrusion detection architecture enhancing trust based approaches’, First International Workshop on Wireless Information System, 4th International Conference on Enterprise Information System, Universidad de Castilla-La Mancha Ciudad Real, Spain. Ali, A. and Fisal, N. (2008a) ‘A real-time routing protocol with load distribution in wireless sensor networks’, Computer Communications, Vol. 31, pp.3190–3203. Ali, A. and Fisal, N. (2008b) ‘Security enhancement for real-time routing protocol in wireless sensor networks’, 5th IFIP International Conference on Wireless and Optical Communications Networks, WOCN ‘08. Ali, A., Latiff, L.A. and Fisal, N. (2009) ‘Simulation-based real-time routing protocol with load distribution in wireless sensor networks’, Wireless Communications and Mobile Computing, 9999, n/a.
Biological inspired secure autonomous routing mechanism for WSN
23
Ali, A., Latiff, L.A., Sarijari, M.A. and Fisal, N. (2008) ‘Real-time routing in wireless sensor networks’, The 28th International Conference on Distributed Computing Systems Workshops, IEEE, Beijing, China. Al-Karak, J.N. and Kamal, A.E. (2004) ‘Routing techniques in wireless sensor networks: a survey’, Wireless Communications IEEE Journal, Vol. 11, pp.6–28. Alpcan, T. and Basar, T. (2003) ‘A game theoretic approach to decision and analysis in network intrusion detection’, Proceedings 42nd IEEE Conference on Decision and Control, 2003, IEEE, The Hyatt Regency Resort & Spa. Maui, Hawaii, USA. Avizienis, A. (1997) ‘Toward systematic design of fault-tolerant systems’, Computer, Vol. 30, pp.51–58. Balasubramaniam, S., Botvich, D., Donnelly, W., Foghluh, M. and Strassner, J. (2006a) ‘Biologically inspired self-governance and self-organisation for autonomic networks’, Proceedings of the 1st International Conference on Bio Inspired Models of Network, Information and Computing Systems, ACM, Cavalese, Italy. Balasubramaniam, S., Donnelly, W., Botvich, D., Agoulmine, N. and Strassner, J. (2006b) ‘Towards integrating principles of molecular biology for autonomic network management’, Hewlett Packard university Association (HPOVUA) Conference, Nice, France. Banerjee, S., Grosan, C. and Abraham, A. (2005a) ‘IDEAS: intrusion detection based on emotional ants for sensors’, in IEEE (Ed.): Proceedings of the 2005 5th International Conference on Intelligent Systems Design and Applications (ISDA ‘05), Wroclaw, Poland. Banerjee, S., Groşan, C., Abraham, A. and Mahanti, P.K. (2005b) ‘Intrusion detection on sensor networks using emotional ants’, International Journal of Applied Science and Computations, Vol. 12, pp.152–173, USA. Boonma, P. and Suzuki, J. (2008) ‘MONSOON: a coevolutionary multiobjective adaptation framework for dynamic wireless sensor networks’, Proc. of the 41st Hawaii International Conference on System Sciences (HICSS), Big Island, HI. Cerpa, A., Wong, J.L., Kuang, L., Potkonjak, M. and Estrin, D. (2005) Statistical Model of Lossy Links in Wireless Sensor Networks, ACM/IEEE IPSN, Los Angeles, USA. Chen, M., Kwon, T., Yuan, Y. and Leung, V.C.M. (2006) ‘Mobile agent based wireless sensor networks’, Journal of Computers, Vol. 1. Chen, Y. and Nasser, N. (2006) ‘Energy-balancing multipath routing protocol for wireless sensor networks’, The Third International Conference on Quality of Service in Heterogeneous Wired/Wireless Networks (QShine ‘06), Waterloo, Ontario, Canada, ACM. Chowdary, T.H. (2002) ‘Telecommunications in developing countries’, Int. J. Inf. Technol. Manage., Vol. 1, pp.325–344. Das, A.A.D.K. and Basu, K. (2004) ‘A game theory based approach for security in wireless sensor networks’, IEEE International Conference on Performance, Computing, and Communications, 2004, IEEE, Phoenix, Arizona. de Lemos, R., Timmis, J., Ayara, M. and Forrest, S. (2007) ‘Immune-inspired adaptable error detection for automated teller machines’, Systems, Man, and Cybernetics, Part C: Applications and Reviews, IEEE Transactions on, Vol. 37, pp.873–886. Dressler, F. (2006) ‘Benefits of bio-inspired technologies for networked embedded systems: an overview’, Dagstuhl Seminar Proceedings 06031, Organic Computing – Controlled Emergence, Dagstuhl, Southwest Germany. Du, X., Xiao, Y., Chen, H.H. and Wu, Q. (2006) ‘Secure cell relay routing protocol for sensor networks: research articles’, Wireless Communications and Mobile Computing, Special Issue: Wireless Network Security, Vol. 6, pp.375–391. Felemban, E., Lee, C.G., Ekici, E., Boder, R. and Vural, S. (2005) ‘Probabilistic qos guarantee in reliability and timeliness domains in wireless sensor networks’, 24th Annual Joint Conference of the IEEE Computer and Communications Societies, IEEE Proceedings, Miami, FL, USA. Frodigh, M., Johansson, P. and Larsson, P. (2000) ‘Wireless ad hoc networking – the art of networking without a network’, Ericsson Review, Vol. 4, pp.248–263.
24
K. Saleem et al.
Gao, H.H., Yang, H.H. and Wang, X.Y. (2005) ‘Ant colony optimization based network intrusion feature selection and detection’, Proceedings of 2005 International Conference on Machine Learning and Cybernetics, 2005, Guangzhou, China. Harmer, P.K., Williams, P.D., Gunsch, G.H. and Lamont, G.B. (2002) ‘An artificial immune system architecture for computer security applications’, IEEE Transactions on Evolutionary Computation, Vol. 6, pp.252–280. Jiang, Y-C., Zhong, Y-P. and Zhang, S-Y. (2005) ‘The agent migration and fault-tolerance management in dynamic network security system’, Int. J. Inf. Technol. Manage., Vol. 4, pp.94–112. Kachirski, O. and Guha, R. (2003) ‘Elective intrusion detection using multiple sensors in wireless ad hoc networks’, Proceedings of the 36th Annual Hawaii International Conference on System Sciences (HICSS ‘03), Hawaii, USA. Karlof, C. and Wagner, D. (2003) ‘Secure routing in wireless sensor networks: attacks and countermeasures’, Ad Hoc Networks, Vol. 1, pp.293–315. Lee, C. and Suzuki, J. (2007) ‘Autonomic network applications designed after immunological self-regulatory adaptation’, in IEEE (Ed.): International Conference on Integration of Knowledge Intensive Multi-Agent Systems, KIMAS, Waltham, MA. Lee, C. and Suzuki, J. (2010) ‘SWAT: a decentralized self-healing mechanism for wormhole attacks in wireless sensor networks’, in Xiao, Y., Chen, H. and Li, F. (Eds.): Handbook on Sensor Networks, World Scientific Publishing Co. Li, G., He, J. and Fu, Y. (2008) ‘A distributed intrusion detection scheme for wireless sensor networks’, in IEEE (Ed.): The 28th International Conference on Distributed Computing Systems Workshops, Beijing, China. Lim, C.H. (2008) ‘LEAP++: a robust key establishment scheme for wireless sensor networks’, in IEEE (Ed.): The 28th International Conference on Distributed Computing Systems Workshops, Beijing, China. Ma, J., Zhang, S., Zhong, Y. and Tong, X. (2007) ‘SAID: a self-adaptive intrusion detection system in wireless sensor networks’, Information Security Applications. Mazhar, N. and Farooq, M. (2007) BeeAIS: Artificial Immune System Security for Nature Inspired, MANET Routing Protocol, BeeAdHoc, LNCS 4628, pp.370–381, Springer-Verlag Berlin Heidelberg. Okdem, S. and Karaboga, D. (2006) ‘Routing in wireless sensor networks using ant colony optimization’, in Okdem, I. (Ed.): Proceedings of the First NASA/ESA Conference on Adaptive Hardware and Systems (AHS ‘06), Istanbul. Okdem, S. and Karaboga, D. (2009) ‘Routing in wireless sensor networks using an ant colony optimization (ACO) router chip’, Sensors, Vol. 9, pp.909–921. Pathan, A.K., Lee, H.W. and Hong, C.S. (2006) ‘Security in wireless sensor networks: issues and challenges’, Proceedings of 8th IEEE ICACT 2006, Phoenix Park, Korea. Prattipati, N. and Hart, E. (2008) ‘Evaluation and extension of the AISEC email classification system’, Artificial Immune Systems. Ramadan, R.A. (2009) ‘Agent based multipath routing in wireless sensor networks’, in IEEE (Ed.): Intelligent Agents, 2009, IA ‘09, Nashville, TN. Saleem, K., Fisal, N., Abdullah, M.S., Zulkarmwan, A.B., Hafizah, S. and Kamilah, S. (2009a) ‘Proposed nature inspired self-organized secure autonomous mechanism for WSNs’, in IEEE (Ed.): Asian Conference on Intelligent Information and Database Systems, Quang Binh University, Dong Hoi City, Quang Binh Province, Vietnam. Saleem, K., Fisal, N., Hafizah, S., Kamilah, S. and Rashid, R.A. (2009b) ‘Ant based self-organized routing protocol for wireless sensor networks’, International Journal of Communication Networks and Information Security (IJCNIS), Vol. 2, pp.42–46. Secker, A., Freitas, A.A. and Timmis, J. (2003) ‘AISEC: an artificial immune system for e-mail classification’, The 2003 Congress on Evolutionary Computation, 2003, CEC ‘03, IEEE Press, Canberra, Australia.
Biological inspired secure autonomous routing mechanism for WSN
25
Singh, G., Das, S., Pujar, S. and Gosavi, S. (2004) ‘Ant colony algorithms for Steiner trees: an application to routing in sensor networks’, in de Castro, L.N. and von Zuben, F.J. (Eds.): Recent Developments in Biologically Inspired Computing, Idea Group Publishing. Siraj, A.B.R., Susan, V. and Bridges, M. (2004) ‘Intrusion sensor data fusion in an intelligent intrusion detection system architecture’, Proceedings of the 37th Annual Hawaii International Conference on System Sciences (HICSS ‘04), Hawaii. Stuetzle, T. and Dorigo, M. (2002) ‘A short convergence proof for a class of ACO algorithms’, IEEE Transactions on Evolutionary Computation, Vol. 6, pp.358–365. Su, C-C., Chang, K-M., Horng, M-F. and Kuo, Y-H. (2005) ‘The new intrusion prevention and detection approaches for clustering-based sensor networks’, in IEEE (Ed.): Wireless Communications and Networking Conference, New Orleans, USA. Walters, J.P. and Liang, Z. (2007) ‘Wireless sensor network security: a survey’, Security in Distributed Grid and Pervasive Computing, Auerbach Publications, CRC Press. Wang, H., Sheng, B., Tan, C.C. and Li, Q. (2008) ‘Comparing symmetric-key and public-key based security schemes in sensor networks: a case study of user access control’, in IEEE (Ed.): The 28th International Conference on Distributed Computing Systems, Beijing, China. Wen, Y-F., Chen, Y-Q. and Pan, M. (2008) ‘Adaptive ant-based routing in wireless sensor networks using energy*delay metrics’, Journal of Zhejiang University SCIENCE A, Vol. 9, pp.531–538. Wood, A.D. and Stankovic, J.A. (2002) ‘Denial of service in sensor networks’, Computer, Vol. 35, No. 10, pp.54–62. Xing, K. and Liu, F. (2008) ‘Real-time detection of clone attacks in wireless sensor networks’, in IEEE (Ed.): The 28th International Conference on Distributed Computing Systems, Beijing, China. Zhao, J. and Govindan, R. (2003) ‘Understanding packet delivery performance in dense wireless sensor networks’, Proceedings of the 1st International Conference on Embedded Networked Sensor Systems, Los Angeles, USA.
