Secure Data Aggregation Based on End-to-End Homomorphic Encryption in IoT-Based Wireless Sensor Networks

The IoT-based wireless Sensor network (WSN) is a revolutionary system for smart observation. An IoT-based Wireless Sensor Network (WSN) is defined as a number of spatially dispersed and dedicated sensors for observing and recording the physical conditions, such as temperature, humidity, etc., of the environment. The collected data are forwarded through a wireless network to an internet-based base station. The primary goal of data fusion or aggregation is to extend network life by reducing sensor network resource use, which includes batteries, power, and bandwidth [1,2]. Data aggregation techniques, on the other hand, could affect key quality of service measures in WSN, such as accuracy, speed, and failure [3]. Furthermore, data aggregation introduces new risks. A hacked sensor node, for instance, might either fraudulently release or broadcast the data it acquires from neighbouring nodes, or return random results as aggregated data. As a result, an opponent may violate both the secrecy and the integrity of the information over a broad section of the WSN by compromising a significant number of aggregating units near the base station. Rafik et al. [4] offered a secure data accumulation strategy which guarantees data privacy through symmetric-key homomorphic encryption (HE) using homomorphic signatures to validate data integrity. The protocol [4] is prone to wormhole attack. Lacking understanding about the key procedures of cryptography, a wormhole exploits the network communication architecture. Wormhole threats are primarily designed to confuse routing protocols and communication services [5]. As a result, developing an end-to-end secure data aggregation technique, while achieving confidentiality, integrity and attack detection, is a challenging task because an effective security strategy is essential in order to preserve integrity and durability and to retain sensitive data. The key objectives of the proposed protocol are as follows.

The literature discusses a vast scope of secure data aggregation methodologies centered around homomorphically encrypted algorithms. Hung et al. suggested a solution [6] that guarantees the confidentiality of data, reliability, and resistance to eavesdropping threats. Additionally, it recognizes malicious activity with certain added costs. The scheme’s primary issue is that no source authentication is accomplished, making it susceptible to Sybil threats. The above limitation is taken into account in the concealed data aggregation (CDA) method [7] in which data aggregators (DA) perform only gathering and merging operations using encrypted messages. As a result, DAs are not required to hold vulnerable encryption keys. The concept of clustering along with data aggregation was introduced by the authors of the protocol [8], in which the notion of clustering, as well as data aggregation and algebraic features of polynomials, was included. A further end-to-end data aggregating strategy [9], which employs homomorphic encryption includes elliptic curve cryptography. Compared with previous complicated techniques, the elliptic curve cryptography technique allows nodes to produce keys with a reduced key size. This technique is remarkable for generating two distinct encrypted messages given two exactly identical messages. This is robust to documented plain text threats, and man-in-the-middle attacks. However, such a solution merely gives secrecy, neither authentication nor integrity. In order to achieve integrity and authentication along with confidentiality, there was a need for more efficient secure data aggregation techniques. Authors [10] suggested a simple and evidently secure encrypting approach relying upon indistinguishable characteristics of a cryptographic basic, Pseudo Random Function (PRF). The methodology assumes the integrity of aggregated data, but also end-to-end authentication. The fundamental aspect of the next protocol, SEEDA, is that it has minimal network communication overheads. The constraint of such an approach is that it only achieves secrecy and does not ensure integrity or authentication. Another issue, along with the achievement of confidentiality, integrity and authentication, is that the data aggregation process is vulnerable to various attacks, such as false data injection, node compromised attacks, Sybil and wormhole attacks. Data aggregation incorporating fake information monitoring is enabled by a protocol [11]. To achieve this, the monitored nodes from every Data Aggregator (DA) not only perform data aggregation but also calculate the appropriate MAC for data verification at respective allocated paired members during the design stage. To ensure data exchange confidentiality, the sensor nodes in between subsequent DAs verify the data integrity of the encrypted information instead of the raw text. The technique [12] was originally designed for multilayer data gathering and employs various keys for data encryption. For data integrity and secrecy, it utilizes elliptic curve cryptography and HE. This is resilient to replay vulnerabilities, eavesdropping, recognized simple text attacks, encrypted analysis, and illegal aggregation. Energy Efficient Hierarchical Aggregation (EEHA) achieved confidentiality and integrity, but it only addressed eavesdropping and replay attacks [13]. SEDA-ECC [14] achieves only confidentiality with the help of ECC (Elliptic Curve Cryptography), but the main focus was on the compromised node attack. The protocols [15,16,17,18,19,20] achieved confidentiality, authentication and integrity, along with addressing of false data injection attacks, snooping and man in middle attacks with the full HE scheme. Researchers in [21,22,23,24,25,26] have proposed various quality of service (QoS) metrics, but without securing the data transmission that can be blocked in IoT-based WSN networks. In response to waves in the ocean, aquatic creatures, or ships that are passing, sensors in underwater WSN may sway up to 3 m/s. The movement of nodes and networking disconnections have not been taken into account in the node installation and localization strategies previously in operation. To overcome this issue, a dynamic topology control algorithm for node deployment (DTCND) in mobile UWSN was suggested in [27]. For underwater intelligent traffic control and underwater vehicle navigation systems, it is also regarded as a crucial and difficult component. The researcher group in [28] offers an overview of the fundamental ideas behind multisensor data fusion and a thorough analysis of the most widely used data fusion architectural models for marine wireless sensor networks. In [29], different routing protocols—AODV, DSR, and WRP—are analyzed. Two possibilities that depend on the number and movement of nodes have been taken into consideration. PDR and throughput both improve as the number of nodes increases. Furthermore, less delay was caused by low node density. In both instances, AODV had a greater packet delivery ratio and throughput, while WRP had the least delay. The efficiency of the ad hoc network was determined by the authors’ analysis of the average energy usage and the optimum routing protocol that was determined by the outcome. The researchers used IOT with WSN to recognize and transfer the data [30]. With cloud-assisted healthcare WSNs, a secure and reliable certificateless publicly monitored approach that provides adaptive data exchange and confidentiality control, as well as effective grouped user revoking, was introduced in [31]. Table 1 presents a comparative review of the numerous methodologies suggested for secure data aggregation. The analysis is focused around the following parameters: data confidentiality, data integrity, provider authentication, node reliability, attack prevention, aggregating functionality, and strategies adopted to accomplish secure data aggregation. Thus, according to Table 1, there have been diverse secure data aggregation methods that resolve the challenges of authentication, confidentiality, integrity, and threats such as sensor node compromised attacks, misleading data injectors, replay attacks, spoofing, defined simple text threats, encrypted assessment, and unauthorized access collation. However, none of the existing protocols addressed the problem of wormhole attacks in data aggregation. As a result, a novel data aggregation architecture that enables end-to-end data protection, verification, durability, and includes wormhole attack monitoring, is needed. The novel strategy must be reliable and effective in respect to power utilization, propagation latency and aggregation output accuracy. The proposed protocol in this research is an improvement over the existing protocol [4]. The methodology of Othman et al. [4] is founded around symmetric key HE to ensure data privacy, as well as applying a homomorphic signature to validate the aggregated data integrity. This could efficiently ensure data privacy, validate data, and obtain greater transmission performance. The proposed protocol is based on an asymmetric key HE, and asymmetric key techniques are more efficient in terms of key distribution, security level, and security services as compared with symmetric key techniques [5]. In an asymmetric key technique, integrity, authentication and non-repudiation are achieved along with data confidentiality. The existing protocol [4] is vulnerable to wormhole attack, as wormhole attack does not depend on strong cryptographic keys. Wormhole attacks can be carried out by a solitary or perhaps a couple of cooperating nodes, as well as by two or even more intruders associated by a high channel link known as a wormhole link.

The protocol proposed in this paper uses end-to-end encryption, namely, EEHE based on asymmetric primitives. In this article, we first offer an brief overview of each of these aspects, along with additional cryptographic techniques, prior to discussing our modeling techniques and architectural aims.

The proposed protocol (as illustrated in Proposed Algorithm 1) is based on the concept of a group of three sensor nodes designated as the monitoring node, neighbouring node and forwarding node. The group is called the MNF group. The flow of the proposed work can be understood from Figure 3 and the timing diagram shown in Figure 4. Detailed steps are as follows.

In the proposed method, the aggregated data meet the security criteria of secrecy, data integrity, wormhole attack detection, and false data identification.

To measure the recommended method’s performance, it is simulated utilizing OMNET++, and its performance across varying conditions is analyzed. Quantity matrices comprising mean power dissipation, network lifespan, and the proportion of surviving nodes are used to evaluate performance. Across the simulation, a network topology of 100 nodes is considered, with every node possessing an initial energy of 2 Joules. In the first simulation, a comparison of the proposed protocol with an existing protocol [3,4], in terms of the probability of false positives and probability of wormhole detection, is performed. In the second simulation, a comparison of the proposed protocol’s average energy dissipation and system lifetime with the existing protocol is performed.

In this paper, a technique is described that ensures confidentiality in data fusion between a transmitter and receiver and also identifies wormhole threats on a specific node utilizing the shared neighbours table. In the proposed protocol, a common neighbour table is created in order to perform an integrity check and detect a wormhole. Due to the overhearing property of wireless communication, common neighbours between a sender and a receiver decide whether the receiver is a wormhole or not. Thus, there are fewer chances of false positive results. The identification and removal of all wormhole nodes in the network are also discussed in the proposed protocol. An EEHE technique is used for achieving end-to-end confidentiality of the aggregated data. The proposed scheme requires less energy compared with the protocol [3,4], and there are fewer chances of false positives. The key features of the SDT technique are that it does not require any guard nodes or special hardware, and there is no data loss owing to wormhole events. Whenever a false node is discovered, an alert alarm is triggered without any additional directional antennas. In the present work, neighbouring information, particularly node ID, is used to detect the presence of wormholes. This scheme fails whenever there exists a malicious node with a fake node ID which acts as a trustworthy node and misleads the routing protocol. In future, the location-based wormhole attack detection protocol could be taken into account.