Projects

Abstract— Access control, the mechanism by which computer systems determine who is authorized to access specific resources, has evolved significantly over time. Early approaches relied on simple access control lists and role-based models, which were sufficient for small-scale or isolated systems. However, as systems have grown in scale and complexity, and as connectivity between users and resources has increased, more expressive and flexible models have become necessary. One such model is Relationship-Based Access Control (ReBAC), which determines permissions based on the relationships between users, resources, and groups. Systems like Google's Zanzibar have demonstrated how ReBAC can be applied effectively at large scale, enabling fine-grained access control for global online services. Despite its advantages, ReBAC introduces several technical challenges, particularly when implemented over distributed, large-scale graph structures. These include performance bottlenecks during permission checks, maintaining consistency across distributed components, and efficiently handling frequent updates to the underlying relationship graphs. In this paper, we explore the design, implementation and optimization of authorization systems based on relationship graphs. We examine key challenges and investigate potential solutions, including advanced indexing techniques, caching techniques sampling and Machine Learning based algorithms for graph traversal to improve authorization performance.

Abstract— The pharmaceutical industry in India grapples with challenges related to the unauthorized procurement of controlled drugs listed in Schedule H and beyond, often due to the absence of genuine prescriptions or the use of invalid ones. In response, this paper proposes a targeted permissioned blockchain-based solution to ensure the authenticity and validity of prescriptions for such controlled medication. By integrating blockchain technology, the system provides a secure and transparent platform for issuing prescriptions by authentic doctors and verifying them by authorized pharmacists. The inclusion of nodal authorities, responsible for implementing the Drug and Cosmetic Act and its rules, as validators in the Proof of Authority (PoA) consensus incentivizes them to adopt this model. By employing PoA, the benefits of blockchain as a decentralized ledger can be utilized, along with fast transaction approvals and high scalability. Our approach not only mitigates the risks associated with privacy concerns of patient information but also enhances accountability and trust within the healthcare ecosystem. The paper surveys the present research regarding application of blockchain based prescription systems across the world, and discusses the feasibility of such a system in the Indian context. 

Abstract—  Multiple Cloud Compute providers can be managed together as a Federated Cloud. Such an agglomeration introduces challenges for effective resource allocation and load balancing. To address such issues, multiple papers have been written describing various algorithms for resource allocation. This paper thus compares and contrasts various algorithms proposed. A Federated Cloud environment is usually organized as a many providers having multiple datacenters (DCs), each with multiple hosts, and each host having multiple VMs. We compare the various algorithms on the basis of execution time (time-till-response), latency (time between task scheduling and VM allocation), and load (resource utilization).