Research

Today, most vehicles on the road like cars, trucks, and motorcycles have internet connectivity. This internet connectivity allows cars to access convenient features like hands-free calling, emergency service access, and weather reports. With improved short-range communications, these vehicles can also form a VANET on the road. However, the ability to connect to the internet and perform inter-vehicular communications makes VANETs and connected vehicles targets for cyber attacks. To protect against cyber attacks in these scenarios, the usage of ML can be an interesting research area as ML contains the ability to find correlations in network data. Leveraging the power of ML can provide efficient solutions to help forecast and detect cyber attacks before they can cause irreversible damage.

The usage of ML for various applications has gained attention due to its ability to find correlations and forecast predictions effectively. However, the increased usage of ML has made these frameworks an essential cyber threat from adversarial attacks. Adversarial attacks are cyber attacks that aim to corrupt the functionality of ML algorithms by performing adversarial manipulations on them. These attacks aim to manipulate training data and model sensitivities to adversely affect the performance of the classifier.  Therefore, the development of mechanisms to detect and thwart adversarial attacks is an interesting open research area. This is specifically necessary in scenarios like Smart Cities and Autonomous Driving.

Today's cyber-physical systems are an integral part of modern computing environments. These architectures combine the physical environment with cyber technology to create infrastructures that are efficient, transparent, and responsive. One instance of a cyber-physical system is the Smart Grid, which helps reduce greenhouse gas emissions and other pollutants by facilitating the connection of large amounts of renewable energy. It also enables technologies that make it easier for customers to reduce energy use or shift energy use to times when prices and emissions are lower. Unfortunately, the Smart Grid has become a popular target for cyber attacks. Through the usage of ML techniques, new research can be conducted to protect Smart Grid and its applications from cyber attacks.  Potential research in ML-enabled Smart Grid security solutions can be beneficial in promoting cyber resilience and awareness in the Smart Grid. 

The modern IoT network is forecasted to grow immensely to 125 billion devices by 2030. The increased usage of IoT technology also increases the potential for cyber attacks across the networking infrastructure. Cyber attacks on IoT devices can not only impact the device/organization/system, but they can also send a ripple effect to other interconnected devices, causing a large-scale system malfunction. To protect against IoT attacks, the usage of ML for securing IoT infrastructures can be an attractive research opportunity. Generating ML solutions for IoT security can help with being able to analyze cyber threats through data-driven mechanisms. This can ensure a more secure cyberspace for the future of IoT technology. 

Network anomalies can be problematic in SDNs as they can be suggestive of malicious network intrusions and cyber attacks like DDoS, Blackhole, and MITM.  These scenarios can cause loss of revenue, reputation, and intellectual property for organizations, and their effects can be exacerbated if the SDN is connected to national critical infrastructures. To protect against network anomalies, the usage of ML for anomaly and intrusion detection is an interesting research opportunity due to the ability of ML to find correlations in network data. Leveraging the power of ML can provide efficient solutions to help SDN administrators forecast and detect cyber attacks before they can cause irreversible damage to the system.