Tree-HP2PL: A Hierarchical Priority-Based Two-Phase Locking Protocol for Nested Transactions in Distributed Real-Time Database Systems

Modern applications require advanced concurrency control protocols to efficiently manage nested transactions within Distributed Real-Time Database Systems (DRTDBS). This paper introduces Tree-HP2PL, a Hierarchical Priority-Based Two-Phase Locking protocol designed to optimize conflict resolution within nested transactions. The proposed method employs a binary search tree structure for transaction management and incorporates a hybrid strategy combining Depth-First Search (DFS) and parallel Breadth-First Search (BFS) for efficient lock acquisition. Tree-HP2PL is capable of distinguishing and resolving various conflict scenarios, including no conflict, intra-transaction conflicts, and inter-transaction conflicts. To mitigate pseudo-priority inversion, it implements priority inheritance and transaction abort mechanisms as necessary. The protocol’s effectiveness was evaluated through experimental simulations conducted in a MATLAB simulation environment, using transaction volumes ranging from 5 to 30 across 10 iterations. Results indicate that execution time remains efficient, with a recorded time of 0.0011 seconds for 5 transactions and showing controlled growth to 0.0049 seconds at 20 transactions. Memory usage demonstrated linear scalability, increasing from 0.006 MB at 5 transactions to 0.016 MB at 30 transactions. Although system throughput decreases under heavier transaction loads, it remains competitive, dropping from 1980 transactions per second (TPS) at 5 transactions to 998 TPS at 30 transactions. Notably, the success ratio consistently maintained a value of 100%, reflecting the protocol’s high reliability. In conclusion, Tree-HP2PL effectively addresses the challenges of concurrency control in nested transaction environments by offering a scalable, structured, and conflict-aware locking mechanism. Its robust performance and methodological design validate its applicability in real-time distributed systems, marking it as a significant improvement over existing flat or unstructured locking techniques.