RA.K. Saravanaguru’s research while affiliated with Vellore Institute of Technology University and other places

Cloud is based on the underlying technology of virtualization. Here, the physical servers are divided into multiple virtual servers. Through the technology of virtualization, each virtual server contains virtual machines. Live virtual machine migration is expected to be with the aim of having migration time and inactive time with minimal duration. Various machine-learning approaches have been investigated and identified research gaps to enhance the security features during the migration process. Moreover, a secure virtual machine live migration is proposed using Advanced Metric Encryption (AME). Considering the duration of live migration in data centers as well as ensuring the security aspects, the proposed model has been tested and evaluated.

Digital collection and maintenance of evidence corresponding to an accident is of utmost relevance as vehicles become connected and part of Internet of things ecosystem. A conceptual evidence management framework is proposed examining the future of accident investigation forensics in the era of connected vehicles. The framework discusses how evidence generated from vehicles involved in incident along with supporting evidence from nearby vehicles CCTVs and road users can be collected and managed over blockchain using smart contracts in a vehicle to everything connected environment. The management of evidence over blockchain serves as an immutable and auditable means to investigate and settle cases post an accident among the stakeholders. Dynamic access control over evidence data and reports based on location of incident and vehicle involved is implemented with the help of smart contracts. The cost of deploying and carrying out transactions using the smart contracts is evaluated over both public and private Ethereum blockchain. Mechanism and cost involved in storing evidence on chain versus off chain using Inter Planetary File System in most optimised manner saving memory and execution cost is compared and presented. The smart contract codes are deployed on Rinkeby test network and can be accessed using address mentioned.

Timely detection of traffic events is of upmost importance in contributing towards traffic safety and ease of commute. A spatial mobile crowdsourcing framework is proposed, enabling road users to report traffic events such as accidents, traffic rule violation, bad road and environment conditions through their mobile devices using images, videos and mobile sensor data. The framework is designed as a decentralized application over blockchain and makes use of interplanetary file system for storage. As transportation infrastructures like CCTV and road side units become increasingly intelligent over time, they are expected to detect traffic events like accidents and traffic rule violations at the edge using machine learning techniques without human intervention. Development and training of traffic event detection models at the edge require training the system using labelled training instances. Crowdsourcing is proposed as a means of collecting training data and validation of event during the learning phase.

Vehicles will become connected via Internet of vehicles as part of Intelligent transportation systems in future. The road users like pedestrians and cyclist can become part of the connected environment using mobile applications. A conceptual framework involving the Convergence of emerging technologies like machine learning, blockchain, mobile application, Internet of vehicles and Edge computing is presented. The work studies how these existing individual technologies can converge together to incorporate safety and evidence management in the context of pedestrian hit and run cases. Blockchain is used as a secure, distributed and immutable forensic log to manage evidence related to the incident by various stakeholders for investigation and settlements. Smart contract specific to storage and access of evidence related to hit and run case is designed and evaluated.

Even though there is continuous improvement in road and vehicle safety, road traffic incidents have been increasing over last few decades. There is a need to reduce traffic incidents like accidents through predictive analysis and timely warnings while at the same time data related to accidents and traffic violations need to be maintained in a tamper proof storage system that can be retrieved for forensic analysis and law enforcement at a later stage. The Secure Incident and Evidence Management Framework (SIEMF) proposed in this work address these two challenges of predictive modeling for timely warning and secure evidence management for forensics analysis in case of accidents and traffic violations. The system proposes a deep learning based predictive incident modeling with blockchain and CP-ABE based access control for the incident data stored in blockchain.