Final Architecture Analysis for the Project:

"BLE-based Indoor Asset and People Tracking"

Architecture Overview:

Edge → Database → GEM → Widget

---

1. Edge-Level Architecture:

We utilized BLE devices to detect the location of assets or people indoors. Specifically, we worked with SmartBond™ DA14695 Bluetooth Low Energy 5.2 Daughter Boards/Main Boards and SmartBond™ Wireless Ranging (WiRa™) technology provided by Renesas.

BLE Gateways:

• Custom Gateways Implementation:

  • Two custom BLE Gateways were implemented using DA14695 Main Boards with the BlueZ protocol stack.

  • The Python pybluez library was used to access BLE data from the Main Board for two distinct demos:

    1. Proximity Sensing Demo:

       • BLE signals were filtered to focus on two specific DA14695 Daughter Boards.
       • RSSI values were used to classify device proximity into three levels: Immediate, Near, and Far.
       • This demo mimicked a BLE Mesh solution where each Gateway acted as a node capable of peer-to-peer communication or relaying messages Over-The-Air (OTA) through intermediate nodes.

    2. Precise Localization Demo:

       • Leveraged WiRa™ for high-accuracy localization.
       • At least three static beacons (DA14695 Daughter Boards) were placed as responders, while one WiRa™ device configured as an initiator calculated distances to the beacons.
       • Using BLE Range DTE data, triangulation was performed to determine the WiRa™ device’s coordinates.
       • Parsed AltBeacon messages were sent to the custom Gateway for further processing, such as normalization of coordinates for use in the project’s UI.

---

2. Cloud Integration:

• Data Transfer to Cloud:

  • PUT requests sent from the edge to the Yodiwo platform using Postman for request formatting in Python.
  • The Yodiwo database was configured with asset types and extra fields depending on the demo:
    • Proximity Sensing Demo: Extra fields included id, name, proximity_level_from_gateway1, and proximity_level_from_gateway2.
    • Localization Demo: Extra fields included xcoordinate and ycoordinate.
  • These fields were regularly updated to ensure the database remained in sync with real-time asset data.

• UI Implementation:

  • Two widgets were used for visualization: an Asset Tracker Widget for the Proximity Sensing Demo and a Maps Widget for the Localization Demo.
  • Data for these widgets was processed through a custom GEM, which automated data retrieval and rendering.

---

3. GEM Workflow:

The GEM was composed of five stages, facilitating the connection between the database and the UI widgets:

1. Data SRC: Collect account-specific data (e.g., OrgId, RestAPIBasePath) for use in subsequent steps.
2. JS Preparation: Process results from the previous stage and prepare them for querying.
3. Query: Perform POST requests to Yodiwo's API endpoint (https://$(config.RestAPIBasePath)/fm/assets/search) to fetch relevant asset data.
4. JS Handling: Write JavaScript code to transform database data into the appropriate format for the widget.
5. Data DST: Map the output variables from the previous stage to the input ports of the widget for final display.

---

Summary:

This project demonstrated the implementation of a BLE-based asset tracking system that utilized edge computing for real-time processing and cloud integration for visualization. The architecture provided a scalable, efficient solution for both proximity sensing and precise localization.