A Highly Accurate Deep Learning Based Approach for Developing Wireless Sensor Network Middleware
ABSTRACT
Despite the popularity of wireless sensor networks (WSNs) in a wide range of applications, security problems associated with them have not been completely resolved. Middleware is generally introduced as an intermediate layer between WSNs and the end user to resolve some limitations, but most of the existing middleware is unable to protect data from malicious and unknown attacks during transmission. This paper introduces a secure wireless sensor network middleware (SWSNM) based on an unsupervised learning technique called generative adversarial network algorithm. SWSNM consists of two networks: a generator (G) network and a discriminator (D) network. The G creates fake data that are similar to the real sample and combines it with real data from the sensors to confuse the attacker. The D contains multi-layers that have the ability to differentiate between real and fake data. The output intended for this algorithm shows an actual interpretation of the data that is securely communicated through the WSN. The framework is implemented in Python with experiments performed using Keras. Results illustrate that SWSNM algorithm improves the accuracy of the data and enhances its security by protecting data from adversaries.
In addition, the SWSNM algorithm consumes signi_cantly less energy, has higher throughput, and lower end-to-end delay when compared with a similar conventional approach.
EXISTING SYSTEM :
In the last decade, wireless sensor networks (WSNs) have been applied in monitoring systems that are capable of controlling and monitoring various indoor premises, agricultural lands, and forest monitoring applications . Foremost issues associated with WSNs are related to network security due to an increase in the usage of these devices. Traditional security algorithms in WSNs have achieved security goals such as base station protection , cryptography , attack detections , and security location and routing. Many researchers have developed significant solutions to address WSNs’ security challenges. The Intrusion Detection System (IDS) is a security management system that monitors all events within a network. IDS is capable of detecting attacks without compromising network security. The anomaly detection types of Intrusion Detection (ID) can detect any abnormal behavior in the online data. Misuse detection is another type of ID, which works on of_ine data and is able to detect known attacks . These sensors introduce massive data for processing and transmission to the base station. Standard security algorithms are not suitable for WSNs due to limitations in power consumption, low memory (storage capacity), communication capabilities and resource constraints in sensors .
The communication and exchange of information between sensors is a critical challenge due to energy consumption in network. This information must be protected against various threats . The networks should be secured by support security properties such as confidentiality, authenticity, availability, and integrity. Standard applied cryptographic algorithms such as signature and encryption/decryption. However, these mechanisms used secret keys that are unsuitable to the large scale of WSNs due to the large memory requirement to store these keys. Most of these sensors lack physical protection, which leads to compromised nodes. Compromising one or more nodes in a network allows the adversary to launch different attacks to disrupt inter-network communication .
There are various attacks such as adversary, compromised node(s), eavesdropper, etc. These attacks are capable of dropping packets or modify them, resulting in an impact in the performance of WSNs. Source location privacy (SLPs) are mechanisms that protect sensor data from attacks by generating fake nodes. The fake node and packets (dummy message) create fake identity and packets without mentioning the source and destination identity.
DISADVNTAGES :
The drawback of this technique is that it requires more energy and overhead.
PROPOSED SYSTEM :
This paper introduces a secure wireless sensor network middleware (SWSNM) based on an unsupervised learning technique called generative adversarial network algorithm. SWSNM consists of two networks: a generator (G) network and a discriminator (D) network. The G creates fake data that are similar to the real sample and combines it with real data from the sensors to confuse the attacker. The D contains multi-layers that have the ability to differentiate between real and fake data. The output intended for this algorithm shows an actual interpretation of the data that is securely communicated through the WSN. The framework is implemented in Python with experiments performed using Keras.
CONCLUSION :
Wireless sensor networks (WSNs) are an essential medium for the transmission of data for numerous applications. In order to address power consumption, communication, and security challenges, middleware bridges the gap between applications and WSNs. Most existing middleware does not completely address the issues that signi_cantly impact WSNs’ performance. Thus, our paper proposes unsupervised learning for the development of WSNs middleware to provide end-to-end secure system. The proposed algorithm (SWSNM) consists of a generator (G) and a discriminator (D).
The generator is capable of creating fake data to confuse the attacker and resolving imbalanced data by generating more data to balance the proportion of classes, the normal and attack data. We render the D to be a powerful network that can easily distinguish between two datasets, even if the fake data is very close to real samples. Extensive testing on the NSL-KDD dataset with different supervised learning techniques and comparisons with our generator network data shows that our generator model provides a better accuracy of 86.5% with a low FPR of 21% and 84% with a lower FPR of 13% by using full (40 features) and reduced (20 features) respectively. Additionally, we employed the t-SNE algorithm for both full features and reduced features to compare the output of our generator to the original dataset.