A Cloud-Deployed Intelligent Ride Allocation and Dynamic Fare Optimization System Using Python and RESTful Microservices

Abstract: The rapid proliferation of on-demand urban mobility services has underscored the critical need for efficient, scalable, and cost-transparent platforms that can seamlessly connect passengers with nearby drivers in real time. This paper presents the design, implementation, and evaluation of an intelligent cloud-deployable ride allocation and dynamic fare optimization platform developed entirely in Python. The proposed system leverages the FastAPI asynchronous web framework for high-throughput request handling, SQLAlchemy Object-Relational Mapping (ORM) for persistent data management, and OpenStreetMap Nomination for cost-free geocoding augmented by database-level caching. The core algorithmic contributions encompass a Haversine-based greedy nearest-driver matching algorithm and a five-tier rule- based surge pricing model that continuously evaluates real-time demand-to-supply ratios to determine an appropriate fare multiplier. The system supports three role-segregated user classes-passenger, driver, and administrator-governed by session-based authentication reinforced through Cross-Site Request Forgery (CSRF) token validation. An automated test suite comprising 40 scenarios across seven functional modules achieves a 100% pass rate, validating fare computation accuracy, driver assignment correctness, and end-to-end ride lifecycle integrity. Deployment configurations targeting Amazon Web Services Elastic Beanstalk confirm cloud portability. Experimental results demonstrate sub-millisecond fare computation, sub-5-millisecond driver matching, and an 85% geocoding cache hit rate, collectively affirming the platform's suitability for production-grade urban ride-sharing deployment.

Keywords: ride-sharing systems; dynamic fare optimization; Haversine algorithm; nearest-driver matching; surge pricing; FastAPI; cloud deployment; OpenStreetMap Nomination.