- Research Article
- Open Access
A Forward Authentication Key Management Scheme for Heterogeneous Sensor Networks
© Jen-Yan Huang et al. 2011
- Received: 1 June 2010
- Accepted: 23 October 2010
- Published: 25 October 2010
Key encryption technology is a basic technique for protecting the secrecy of transmitted data among sensor nodes in wireless sensor networks. However, sensor nodes are inherently limited by insufficient hardware resources such as memory capacity and battery lifetime. As a result, few current key management schemes are appropriate for wireless sensor networks. This paper proposes a new key management method that uses dynamic key management schemes for heterogeneous sensor networks. The proposed scheme loads a hash function into the base station, cluster heads, and sensor nodes. The cluster heads and sensor nodes then generate their own keychains to provide forward authentication in case of key changes, security breaches, key changes due to security breaches. The cluster heads and sensor nodes establish pairwise keys to ensure transmission secrecy. The proposed scheme decreases the number of keys required for sensor nodes and cluster heads and is robust to the following attacks: guessing attacks, replay attacks, man-in-the-middle attacks, node capture attacks, and denial-of-service attacks.
- Sensor Network
- Sensor Node
- Wireless Sensor Network
- Cluster Head
- Hash Function
Wireless sensor networks (WSNs) consist of many sensor nodes capable of wireless communication and data collection. In addition to sensor nodes, most WSNs include two other components, which are base station and cluster head.
WSNs are suitable for military applications, environmental monitoring, meteorological data collection, medical information monitoring, and so on. WSNs solve the wiring problem that traditional wired networks face. Wireless sensor nodes have the advantages of small size, easy deployment, and dynamic configuration.
Sensor nodes are limited by insufficient hardware resources, such as memory capacity, battery lifetime, and processor speed. The limitations of memory determine the amount of data to be stored, while battery lifetime determines the life of sensor nodes and slow processors cannot handle complex computations. These problems in turn will influence the efficiency of sensor networks.
Researchers have previously proposed some key management schemes for homogeneous sensor networks . In this type of environment, all sensor nodes have the same characteristics, such as battery lifetime, computation power, and memory capacity. However, this scheme encounters the problems of low transmission speed, limited scalability, and a lack of fault tolerance . Heterogeneous sensor networks (HSNs) can avoid these problems. In HSNs, which include several kinds of sensor nodes, different kinds of sensor nodes have different properties and transmission ranges.
This study proposes a key management system for a heterogeneous sensor network. The members of this network include a minority of powerful high-end sensors (H-sensors), which work as cluster heads, and a majority of low-end sensors (L-sensors). The high-end sensors have more memory, a wider transmission range, longer battery and greater fault tolerance. Low-end sensors represent general sensor nodes.
Regarding the security issues in the wireless sensor network, the encrypting scheme must not increase the load of sensor nodes. If sensor nodes need to perform complex computations for encryption, it would consume the energy of sensor nodes. Hence, the traditional encrypting and decrypting method is not suitable for wireless sensor networks.
In the proposed method, the L-sensors only store a little data at a time. Hence, they only require a little memory to work quickly. H-sensors regularly replace the encrypting key based on the status of the cluster. At the same time, the L-sensors can determine if the new key is legal. This design requires fewer resources to achieve the security of sensor nodes in wireless sensor networks, while ensuring confidentiality, integrity, and availability.
Following this introduction, the structure of this paper is as follows. Section 2 reviews related work. Section 3 describes the proposed scheme. Section 4 provides the security analysis. Section 5 presents system analysis. Finally, Section 6 offers conclusions.
This section discusses related research about the foundation of security mechanisms and key management schemes for wireless sensor networks.
2.1. Foundation of Security Mechanism
A typical WSN transmits data between nodes via radio. To protect the security of data transmission, a key cryptosystem can ensure the confidentiality, availability, and integrity of data.
2.1.1. Message Authentication Code
2.1.2. Hash Function
The proposed scheme is based on the one-way hash function [3, 4]. A hash value, generated by a hash function , is given by , in which is a variable-length message and is the hash value with a fixed length. The hash value is appended to the message, allowing the receiver to authenticate it; the hash function itself is not a secret. The hash function is the "fingerprint" of a file, a message, or other block of data.
2.2. Key Management Schemes for Wireless Sensor Networks
In wireless sensor networks, there are three methods of assigning keys: random, deterministic, and hybrid . In the random method, the system randomly chooses several keys from the key pool and then loads them into sensor nodes to create the key-chain. The deterministic method uses dynamic computation to generate keys that can enhance the connection between sensor nodes. In addition, the system can update the key periodically through different situations. By updating the key, the system can isolate malicious nodes and maintain security. The hybrid method combines the advantages of these two methods.
Eschenauer and Gligor  proposed a random key predistribution scheme that focuses on symmetric encryption and decryption. To build the initial encrypting and decrypting key between sensor nodes, the system first generates a huge key pool. The sensor nodes then randomly choose several keys from the key pool and load them before deployment. The sensor nodes use these preloaded keys to generate pairwise keys, which create safe communication channels between neighboring nodes. This communication channel is called a key path, and it allowed sensor nodes to connect with other nodes in the environment. To protect the confidentiality of the key path, each key corresponds to only one index value. However, when an attacker finds the key, the sensor nodes immediately change the index value to update the key and select a new pairwise key.
Chan et al.  proposed a predistribution -composite key method that allows two sensor nodes to set the pairwise key only when they share at least q public keys.
In an attempt to improve upon these two methods, Du et al.  proposed a key management method applicable to heterogeneous sensor networks. This approach uses a small number of sensor nodes that have superior performance to load more keys, increasing the probability of the shared keys.
Liu et al.  proposed the grid-based key predistribution scheme and the random subset assignment scheme. These methods can build the pairwise key between sensor nodes in the wireless sensor networks. Liu and Ning proposed a scheme  that has the great advantage of predicting coordinates in the sensor nodes and then distributing suitable keys in advance.
Li et al.  proposed a hexagonal grid key predistribution scheme that uses a hexagonal coordinate system and binary polynomial. Zhang et al.  proposed a method in which sensor nodes insert their own coordinates and IDs into the hash function and then generate pairwise keys to communicate with each other. This enhances the relationship between sensor nodes. However, this method lacks an authentication scheme between adjacent sensor nodes.
The researches [11–13] proposed the location-aware deployment model of keys predistribution scheme. This approach divides the environment into several square areas and randomly deploys the sensors in each area. The system can be aware of location of the sensor nodes according to the sensor node's ID.
Liu et al.  proposed a group-based keys predistribution scheme that divides the sensor nodes into groups and scatters them. After deployment, the sensor nodes may suffer from wind force or terrain condition, making it likely that in-group sensor nodes likely become neighbors. Finally, they modeled the deployment distribution as a Gaussian distribution. Building pairwise keys between in-group sensor nodes and cross-group sensor nodes offers several advantages. Hence, they built the pairwise key between sensor nodes in the same group using the in-group key predistribution method and used the cross-group key predistribution method to build the pairwise key between adjacent sensor nodes in the different groups.
Moharrum and Eltoweissy  compared the merits and faults of the dynamic key generation and static key generation methods. Based on this analysis, they proposed a new method called an exclusion basis system (EBS) based on the dynamic key management scheme. Eltoweissy et al.  proposed a localized combinatorial keying (LOCK) method that generates the dynamic key based on the EBS.
Perrig et al. presented two security protocols  for sensor networks, called SNEP and μ TESLA hereafter. SNEP achieves data confidentiality and data authentication, while μ TESLA ensures data integrity. In these structures, each sensor node shares the secret key with the base station. The base station functions as a trusted third party to keep and distribute the secret key. Younis et al.  and Jolly et al.  proposed a scheme much like the dynamic key generation model. The scheme can update and change the key through the certification authority (CA).
Chan and Perrig  proposed a protocol called peer intermediaries for key establishment (PIKE). In this approach, sensor nodes are trusted third parties and manage the key. Guorui et al.  proposed a group-based dynamic key management scheme. This system can update and change the key independently of the base station or cluster head. Cheng and Agrawal  proposed an effective method to build and manage the pairwise key. In the scheme, the system generates a two-dimensional key matrix, and each sensor node randomly stores one column and row of the key array from the matrix before deployment. After the sensor nodes are deployed, two adjacent sensor nodes can generate the pairwise key of each other.
Kausar et al.  proposed a hierarchical sensor network consisting of a small number of high-end sensors (H-sensor node) and a large number of low-end sensors (L-sensor node). The scheme is a scalable protocol for key management in the sensor networks to address the sensor nodes resource constraints, including computation, storage, and communication.
This paper proposes a key-chain protocol for key management that is designed for heterogeneous sensor networks (HSNs). Each cluster head generates its own key-chain, which encrypts messages and communicates with the other sensor nodes in the cluster. Based on hierarchical clustering, each cluster consists of several sensor nodes and a cluster head. Several clusters and a base station form the heterogeneous sensor networks.
3.1. System Setup
This section discusses the initialization and authentication phases in HSNs, including setting up the key-chain and setting up pairwise keys for the L-sensor nodes.
H-sensors can directly communicate with the BS.
The base station exchanges messages with L-sensors through H-sensors and vice versa.
H-sensors can send messages to specific L-sensors in the cluster.
H-sensors can broadcast messages to all L-sensors in the cluster.
L-sensors must exchange the messages with each other through an H-sensor. In other words, L-sensors cannot directly exchange messages with each other. Hence, a compromised L-sensor cannot affect the other L-sensor in the cluster.
These communication rules are usually assumed for the hierarchical sensor networks such as SPINS , Gupta and Younis , and LEACH . In this paper, these communication rules should be followed in order to avoid a compromised node infringing the other L-sensors and to prevent the attacks such as replay attacks or man-in-the-middle attacks.
3.1.1. Initialization Phase
Hence, each H-sensor will obtain distinct key-chains, , and random numbers from the BS. H-sensor and L-sensor are stored with the same hash function and , where is a temporary session key for all H-sensors and L-sensors, and . All keys and parameters for each node will be passed from BS to sensor nodes through an offline secure channel.
The definition of the notations.
A temporary session key for all H-sensors and L-sensors
A random number generated by H-sensor
A concatenation operator
3.1.2. Authentication Phase
- (2)The L-sensor may receive one or more hello messages if no barricades are sheltering it. The L-sensor chooses an H-sensor as its cluster head according to the distance and best signal strength of the message. In this environment, each L-sensor notes other H-sensors from which it receives the hello messages, and these H-sensors are recorded as backup cluster heads in case the chief cluster head is disabled. If the L-sensor receives the message, it then takes its own and and generates a pairwise key , replying to the H-sensor. The format of this response message is as follows:
After receiving the response message and of the L-sensor , the H-sensor generates pairwise key . If the condition is satisfied, the H-sensor confirms the validity of the L-sensor ; if not, H-sensor discards the response message. Hence, the H-sensor can use this pairwise key to announce the message or new key of the key-chain to the L-sensor in the cluster.
- (5)After determining all the clustering nodes, the H-sensor broadcasts the ID of members to all the nodes using . If the H-sensor receives the response message from node and node simultaneously, the H-sensor judges whether node and node are neighbors based on the locations. However, this method does not always produce accurate results. If there is a barricade between node and node , it does not have an effect on the security. After judging whether the L-sensors are adjacent, the H-sensor sends all the L-sensor's IDs to the nodes. The format of neighbor message is as follows:
3.2. Normal Operations of HSNs
In the proposed system, the BS generates a key-chain for broadcasting and encrypting messages to the H-sensors. This process is very similar to what the H-sensor does for the L-sensor, as described in Sections 3.1.1 and 3.1.2. To simplify the description of the system structure, this paper omits the details of these procedures. This paper assumes that the BS has generated a key-chain and used the key, say , and pairwise key (the same as cluster key) for all the H-sensors.
This section discusses two different scenarios for the normal operations of the HSNs. Scenario 1 is that the BS broadcasts a message to all the H-sensors to gather the data from all the L-sensors. Scenario 2 is that the BS asks the H-sensor to request the data from the specific L-sensor .
3.3. Adaptability of the Proposed Method
This section discusses the adaptability of the proposed method, including key revocation, addition of a new node, and the generation of a new key-chain.
3.3.1. Key Revocation
3.3.2. Addition of a New Node
After generating the , the H-sensor uses it to send the in a message to L-sensor , where and are the current random number and key in the key chain used by the H-sensor, respectively. The new L-sensor verifies if is satisfied. If yes, then L-sensor confirms the validity of the key to and H-sensor . Otherwise, the L-sensor discards its message, and will select another H-sensor. Finally, L-sensor then transmits the message to the H-sensor using , and then H-sensor broadcasts the neighbor message to all the members once again.
3.3.3. Generation of a New Key-Chain
This section classifies all potential attacks into five categories, such as guessing attacks, replay attacks, man-in-the-middle attacks, node capture attacks, and denial of service attack.
4.1. Guessing Attacks
Guessing attacks are a crucial concern in any security-based system. Assume that an adversary can obtain information or data related to the in the HSNs. Based on this public information, it may be able to guess the . However, the H-sensor will change the to at regular intervals. Further, each L-sensor node can use the pairwise key to encrypt messages to the H-sensor. Therefore, the guessing attack does not have any effect in this environment.
4.2. Replay Attacks
The L-sensor transmits the following message to H-sensor . The message includes H-sensor's ID, plain text , MAC, L-sensor's ID, and a count . When L-sensor delivers the message it increases the once. After the H-sensor receives the message from the L-sensor , it checks the value of to determine if the node suffers from the replay attack.
4.3. Man-in-the-Middle Attacks
Man-in-the-middle attacks are a type of eavesdropping in which the adversary makes independent connections with the nodes and takes over the handling of messages between an L-sensor and the H-sensor. This attack fools sensors into thinking that they are communicating directly with each other over a private connection, when in fact all the details are controlled by the adversary. Based on the rules of the communication between nodes, the L-sensor and the H-sensor use a pairwise key or group key to securely and directly transmit messages to each other (as do the H-sensor and the base station). Therefore, if an adversary does not have the pairwise key or group key, it still cannot eavesdrop or modify the content of the message. Therefore, the man-in-the-middle attack does not have any effect on HSNs.
4.4. Node Capture Attacks
It is difficult to prevent this type of attacks if nodes are not tamper-proof and the environment is unattended. Hence, after all the L-sensors are deployed in the environment, the attacker might acquire some material of the and from the L-sensor using node capture attack. However, the is used twice in authentication phase and is discarded after the establishment of a pairwise key. In our scheme, each L-sensor has a different pairwise key in the cluster. Therefore, based on the property of pairwise keys, if the L-sensor is captured by the adversary and it can gain the interior material of L-sensor , it still cannot obtain the interior material of L-sensor and cannot infect others.
4.5. Denial-of-Service Attacks
Denial-of-service attacks are common attacks in networks, where communication channel in HSNs is public. However, this type of attacks can be detected by enabling the network with an intrusion detection system. The proposed scheme provides protection against this attack. This is because it uses a one-way hash function and MAC in which the H-sensor sends message without expecting any acknowledgement. If the adversary prevents the message from reaching the nodes, neither the H-sensor nor the L-sensor will know about it.
This paper analyzed the proposed method from the following three issues: (1) the number of messages for grouping and establishing the pairwise key; (2) the key sizes; (3) the power consumption analysis. The H-sensors and L-sensors are randomly deployed in 500-square-meter wireless sensor network. This HSN has two types of sensors: a few powerful H-sensor nodes and many L-sensor nodes. The ratio between these two types of sensors is 1 : 10. In our experiments, there are 25 H-sensor nodes and 250 L-sensor nodes. The H-sensor nodes have a key-chain length of 50 keys. The L-sensors are ordinary sensor nodes that are limited in terms of processing capability, power, and memory. They acquire data from the surrounding environment and forward it to the H-sensor nodes. The H-sensor nodes then transmit the data to the base station.
5.1. The Number of Messages between the H-Sensor and L-Sensor
Number of messages for compared methods.
Pairwise key establishment
Kausar et al. 
5.2. The Key Sizes
The number of keys and functions stored in each member of HSNs.
5.3. Power Consumption Analysis
In the simulation, the proposed scheme is compared with the normal HSN without key management rather than the other key management schemes. This is due to the lack of power consumption evaluation in other HSN key management schemes. The power consumption of the proposed scheme was evaluated in terms of the number of survival nodes over rounds. Each round in the simulation is defined as the completion of one of the following three tasks: (1) H-sensor requests and receives the data from all the L-sensors in the cluster; (2) H-sensors requests and receives the data from a specific L-sensor in the cluster; (3) key revocation. For normal HSN without key management, only the first two operations are possible, and the packet length of 46 bytes is also assumed.
The experimental results are shown in Figure 7. The first sensor node that ran out of power occurred at the 428th round in the proposed scheme in contrast to the 579th round for the normal HSN without key management. The whole network died at about the 557th round and the 786th round for the proposed scheme and the normal HSN without key management, respectively. As a result, the proposed method incurred about 29% overhead due to the inclusion of key management scheme. But considering the benefits of the proposed scheme, which include protections against the guessing attacks, replay attacks, and man-in-the-middle attacks as discussed in Section 4, we think the overhead is acceptable and the results could be a starting point for evaluating power consumption on sensor networks with key management.
This study proposes a new key management scheme that is suitable for HSNs. By clustering all the sensor nodes in the environment, cluster heads can generate their own key-chain. The sensor nodes and their cluster heads can jointly establish pairwise keys. Pairwise keys ensure transmission secrecy for each message, protecting data integrity and determining if the sensor nodes are malicious. The key-chain consists of continuous keys, and each key is dependent. This makes it possible for the sensor node to confirm the validity of each key. Sensor nodes or cluster heads through the characteristic of key-chain, when the cluster heads change the key, and then sensor nodes can confirm the identity of the cluster head and the validity of new key. In our scheme, the key is calculated by hash function. The hash function makes it possible to compress data into a fixed length and avoid data collision. Sensor nodes only need to store a few keys and a hash function at a time, reducing the memory requirements of sensor nodes and ensuring key security.
This work was supported partially by National Science Council, Taiwan under Grant NSC 98-2221-E-005-083.
- Kejie L, Yi Q, Jiankun H: A framework for distributed key management schemes in heterogeneous wireless sensor networks. Proceedings of the 25th IEEE International Performance, Computing, and Communications Conference (IPCCC '06), April 2006 513-519.Google Scholar
- Liu D, Ning P, Rongfang LI: Establishing pairwise keys in distributed sensor networks. ACM Transactions on Information and System Security 2005, 8(1):41-77. 10.1145/1053283.1053287View ArticleGoogle Scholar
- Knuth DE: The Art of Computer Programming. 2nd edition. Addison-Wesley, Reading, Mass, USA; 1981.MATHGoogle Scholar
- Nechvatal J: Public key cryptography. In Contemporary Cryptology: The Science of Information Integrity. Edited by: Simmons G. IEEE Press, Piscataway, NJ, USA; 1992.Google Scholar
- Eschenauer L, Gligor VD: A key-management scheme for distributed sensor networks. Proceedings of the 9th ACM Conference on Computer and Communications Security (CCS '02), November 2002 41-47.View ArticleGoogle Scholar
- Chan H, Perrig A, Song D: Random key predistribution schemes for sensor networks. Proceedings of the IEEE Symposium on Security And Privacy, May 2003 197-213.Google Scholar
- Du X, Xiao Y, Guizani M, Chen H-H: An effective key management scheme for heterogeneous sensor networks. Ad Hoc Networks 2007, 5(1):24-34. 10.1016/j.adhoc.2006.05.012View ArticleGoogle Scholar
- Liu D, Ning P: Location-based pairwise key establishments for static sensor networks. Proceedings of the 1st ACM Workshop on Security of Ad Hoc and Sensor Networks, October 2003 72-82.View ArticleGoogle Scholar
- Li G, He J, Fu Y: Key management in sensor networks. Proceedings of the International Conference on Wireless Algorithms, Systems and Applications, August 2006 457-466.View ArticleGoogle Scholar
- Zhang Y, Liu W, Lou W, Fang Y: Securing sensor networks with location-based keys. Proceedings of the IEEE Wireless Communications and Networking Conference (WCNC '05), March 2005 1909-1914.Google Scholar
- Wadaa A, Olariu S, Wilson L, Eltoweissy M: Scalable cryptographic key management in wireless sensor networks. Proceedings of the 24th International Conference on Distributed Computing Systems Workshops, March 2004 796-802.Google Scholar
- Blom R: Non-public key distribution. Proceedings of the International Cryptology Conference on Advances in Cryptology (CRYPTO '98), 1998 231-236.Google Scholar
- Blundo C, Santis AD, Herzberg A, Kutten S, Vaccaro U, Yung M: Perfectly-secure key distribution for dynamic conferences. Proceedings of the 12th Annual International Cryptology Conference on Advances in Cryptology (CRYPTO '93), 1993 471-486.Google Scholar
- Liu D, Ning P, Du W: Group-based key pre-distribution in wireless sensor networks. Proceedings of the 4th ACM Workshop on Wireless Security (WiSe '05), September 2005 11-20.View ArticleGoogle Scholar
- Moharrum MA, Eltoweissy M: A study of static versus dynamic keying schemes in sensor networks. Proceedings of the 2nd ACM International Workshop on Performance Evaluation of Wireless Ad Hoc, Sensor, and Ubiquitous Networks (PE-WASUN '05), October 2005 122-129.View ArticleGoogle Scholar
- Eltoweissy M, Moharrum M, Mukkamala R: Dynamic key management in sensor networks. IEEE Communications Magazine 2006, 44(4):122-130.View ArticleGoogle Scholar
- Perrig A, Szewczyk R, Wen V, Culler D, Tygar JD: SPINS: security protocols for sensor networks. Proceedings of the 7th Annual International Conference on Mobile Computing and Networking, July 2001 189-199.Google Scholar
- Younis MF, Ghumman K, Eltoweissy M: Location-aware combinatorial key management scheme for clustered sensor networks. IEEE Transactions on Parallel and Distributed Systems 2006, 17(8):865-882.View ArticleGoogle Scholar
- Jolly G, Kuscu MC, Kokate P, Younis M: A low-energy key management protocol for wireless sensor networks. Proceedings of the 8th IEEE international Symposium on Computers and Communications, 2003 335-340.Google Scholar
- Chan H, Perrig A: Pike: peer intermediaries for key establishment in sensor networks. Proceedings of the 24th annual joint conference of the IEEE computer and communications societies (INFOCOM '05), 2005 524-535.Google Scholar
- Guorui L, Jingsha H, Yingfang F: A group-based dynamic key management scheme in wireless sensor networks. Proceedings of the 21st International Conference on Advanced Information Networking and Applications Workshops/Symposia, (AINAW '07), May 2007 127-132.Google Scholar
- Cheng Y, Agrawal DP: Efficient pairwise key establishment and management in static wireless sensor networks. Proceedings of the 2nd IEEE International Conference on Mobile Ad-Hoc and Sensor Systems (MASS '05), November 2005 544-550.Google Scholar
- Kausar F, Hussain S, Yang LT, Masood A: Scalable and efficient key management for heterogeneous sensor networks. Journal of Supercomputing 2008, 45(1):44-65. 10.1007/s11227-008-0184-2View ArticleGoogle Scholar
- Gupta G, Younis M: Load-balanced clustering of wireless sensor networks. Proceedings of the International Conference on Communications (ICC '03), May 2003 3: 1848-1852.Google Scholar
- Heinzelman WR, Chandrakasan A, Balakrishnan H: Energy-efficient communication protocol for wireless microsensor networks. Proceedings of the 33rd Annual Hawaii International Conference on System Siences (HICSS '00), January 2000 1-10.Google Scholar
- Savvides A, Han C-C, Strivastava MB: Dynamic fine-grained localization in ad-hoc networks of sensors. Proceedings of the 7th ACM/IEEE Annual International Conference on Mobile Computing and Networking (MobiCom '01), July 2001 166-179.Google Scholar
- Zhang Y, Liu W, Lou W, Fang Y: Location-based compromise-tolerant security mechanisms for wireless sensor networks. IEEE Journal on Selected Areas in Communications 2006, 24(2):247-260.View ArticleGoogle Scholar
This article is published under license to BioMed Central Ltd. This is an open access article distributed under the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.