Skip to content

Changelog

All notable changes to the AI-Driven Predictive Traffic Flow Optimisation System will be documented in this file.

[2.2.0] - 2026-04-10

Added

  • GET /api/traffic/model_info (Java Gateway): New endpoint proxying to Python /model_info. Returns MAPPO model architecture, junction configuration, and agent indices.
  • POST /api/traffic/reset and GET /api/traffic/model_info now require JWT authentication.

Changed

  • RL Inference main.py: Replaced old PPO/Stable-Baselines3 implementation with MAPPO RNNAgent (GRU). Service now uses torch directly — no SB3 dependency.
  • RL Inference requirements.txt: Removed stable-baselines3 and gymnasium; torch==2.2.1 used directly for MAPPO inference.
  • RL Inference Dockerfile: Multi-stage build to reduce image size. CPU-only PyTorch via --extra-index-url https://download.pytorch.org/whl/cpu. Final image ~500 MB (down from 1.35 GB).
  • Java Gateway RlInferenceClient: Added getModelInfo() method proxying to Python /model_info.
  • Config reference: Updated from mappo_sumo_v4.yaml to mappo_sumo_v5.yaml in README and service documentation.
  • index.html: Updated inference engine description from PPO RL Agent (SB3) to MAPPO RL Agent (EPyMARL).

Documentation

  • Java Gateway README.md: Updated protected endpoints list — removed incorrect GET /api/traffic/action, added POST /api/traffic/reset and GET /api/traffic/model_info.
  • Java Gateway README.md: Fixed curl and PowerShell examples for POST /api/traffic/action — added required junctionId field and corrected observation vector to 19 floats.
  • RL Inference README.md: Updated config reference from mappo_sumo_v4.yaml to mappo_sumo_v5.yaml.

[2.1.0] - 2026-04-09

Added

  • JWT Authentication: Implemented stateless token-based security in Java API Gateway
  • HS256 HMAC signing for bearer tokens
  • Configurable token expiration (default 60 minutes)
  • Protected endpoints: /api/traffic/action (GET/POST), /api/traffic/reset
  • Public endpoints: /api/auth/login, /api/traffic/health, Swagger UI

  • New DTOs: LoginRequest, LoginResponse, ErrorResponse

  • LSTM Traffic Predictor Framework: Foundation for time-series forecasting service

  • Architecture defined with FastAPI endpoints (/health, /forecast, /model_info)
  • Data pipeline ready (loader, preprocessor modules)
  • Configured for SUMO integration (edgeData.xml input)
  • Target: 15-minute ahead forecasting with MAE < 10%

  • Supports all 5 major junctions

  • SUMO Documentation: Comprehensive service documentation added

  • docs/sumo/architecture.md — Network structure, file organization, data flow

  • docs/sumo/key-features.md — Traffic flows, route configuration, output formats

  • Enhanced Test Suite: test_api.py updated with improved usability

  • Color-coded output (green/red/blue) for better readability
  • Test result tracking (e.g., "4/4 tests passed")
  • Boolean status returns for each test function

  • Requires: colorama package

  • Documentation Updates:

  • docs/index.md — Added service documentation section, LSTM quick link, security section
  • docs/features.md — Updated feature matrix with JWT checkmarks and LSTM/SUMO status
  • docs/api-gateway/index.md — Added JWT authentication section with config table
  • docs/api-gateway/endpoints.md — Added /api/auth/login endpoint documentation
  • docs/lstm/architecture.md — Complete LSTM service architecture documentation
  • docs/lstm/endpoints.md — LSTM API endpoints (health, model_info, forecast)
  • docs/lstm/key-features.md — LSTM service features and capabilities
  • README.md files updated to reflect JWT integration in main branch

Changed

  • System Architecture Description: Updated to emphasize JWT authentication and LSTM forecasting
  • Feature Matrix: Reorganized to reflect current implementation status
  • JWT moved from "optional" to "implemented"
  • LSTM endpoints marked as framework-ready (awaiting model training)

  • SUMO status clarified: "fully operational as standalone, not yet live-integrated with RL"

  • Java API Gateway README: Enhanced testing section with colorama dependency and color output documentation

Security

  • JWT authentication now production-ready on main branch (previously in A00163691-JWTAuth branch)
  • Stateless token management with configurable expiration
  • Bearer token validation on all protected endpoints

[2.0.0] - 2026-03-31

Changed

  • RL Model: Replaced single-agent PPO (Stable-Baselines3, model.zip) with MAPPO v4 (EPyMARL, agent.th) — shared GRU RNNAgent controlling 5 junctions simultaneously.
  • Python inference service: main.py rewritten to load EPyMARL RNNAgent (PyTorch) instead of SB3 PPO. Architecture: fc1(24→128) → GRUCell(128→128) → fc2(128→4).
  • POST /predict_action payload: Now requires junction_id (string) and obs_data (float[]). Previously only obs_data was needed.
  • POST /api/traffic/action payload (Java): Now requires junctionId (string) and observations (float[], max 19). Previously only observations was needed.
  • Response payload: Both services now include junctionId / junction_id in the prediction response.
  • Environment variable: MODEL_PATH, OBSERVATION_SHAPE_DIM, NUM_AGENTS replaced by MAPPO_AGENT_PATH.
  • Observation dimension: 9 → up to 19 (padded to max junction obs size internally).
  • Agent count: 1 → 5 (one per controlled junction).
  • Docker images: Versioned as rl-inference-service:2.0.0 and traffic-api-gateway:2.0.0.

Added

  • POST /reset_hidden (Python) and POST /api/traffic/reset (Java): Reset MAPPO GRU hidden states between simulation runs.
  • GET /model_info: Now returns full junction map with agent indices and action masks.
  • GRU hidden state management: Per-junction recurrent state maintained between /predict_action calls within a simulation run.

Removed

  • stable-baselines3 and shimmy dependencies from requirements.txt.
  • MODEL_PATH, OBSERVATION_SHAPE_DIM, NUM_AGENTS environment variables.

Performance

  • Mean waiting time reduced 69.8% vs fixed-time baseline (28.8 s → 8.7 s) with MAPPO v4 model.
  • 100% episode completion rate (vs 13.3% for previous IPPO model).

[1.0.0] - 2026-03-21

Added

  • Microservices Architecture: Integrated Java Spring Boot API Gateway and Python RL-Inference service.
  • ML Pipeline: Support for PPO (Proximal Policy Optimization) model inference.
  • Simulation Environment: Integrated SUMO (Simulation of Urban MObility) configurations for the Athlone "Orange Loop".

Changed

  • Monorepo Consolidation: Merged standalone API branch into the main branch to establish a cohesive pipeline.
  • Dependency Management: Updated Python requirements to include FastAPI, Pydantic v2, and Stable-Baselines3.

Security

  • Data Integrity: Initial implementation of TLS 1.3 for telemetry ingestion.
  • Authentication: Added digital signature placeholders in OpenApiConfig.java to prevent data injection attacks.