Enabling Edge Analytics of IoT Data: The Case of LoRaWAN
1 like321 views
The document discusses enabling edge analytics of IoT data using the LoRaWAN network technology. It proposes the IoTRACE framework which augments LoRaWAN's software architecture to support edge computing and analytics of IoT data. This is done by extracting application data at network servers near the edge to enable localized analytics and data services for multiple stakeholders like farmers. A prototype demonstrates initial concepts by emulating LoRaWAN devices and sensors, and performing analytics on the edge using Node-RED flows that access data queues. Future work aims to refine the architectures and data models to support distributed edge-cloud analytics across networks.
Enabling Edge Analytics of IoT Data: The Case of LoRaWAN
- 1. Enabling Edge Analytics of IoT Data: The Case of LoRaWAN Hong-Linh Truong Faculty of Informatics, TU Wien, Austria [email protected] http://rdsea.github.io @linhsolar Global IoT Summit 2018, 4 April 2018, Bilbao 1
- 2. Acknowledgements: Supported by H2020 Inter-IoT (http://www.inter-iot-project.eu/) Tran-Vu Pham, Danh Pham, Phuong Nam Ho and Xuan-Hiep Huynh for discussions on LoRaWAN and aqua/argiculture monitoring in Vietnam Global IoT Summit 2018, 4 April 2018, Bilbao 2
- 3. Outline State of the art, motivating examples and approach IoTRACE key concepts, architecture and main components Prototype and illustrating examples Conclusions and future work Global IoT Summit 2018, 4 April 2018, Bilbao 3
- 4. New network technologies for large-scale monitoring with IoT Large-scale deployment of low-energy sensing devices with long distance communications LoRaWAN, SigFox and NB-IoT Currently two main deployment models Big providers of IoT infrastructures and platform connectivity like Sigfox, telcos or government infrastructures Community with many stakeholders like The Things Network (TTN) But these deployment models are not flexible enough w.r.t IoT data services and edge analytics Global IoT Summit 2018, 4 April 2018, Bilbao 4
- 5. LoRaWAN We focus on LoRaWAN because of its availability and investment cost in developing country context Global IoT Summit 2018, 4 April 2018, Bilbao 5 Suitable for us to work on large-scale geographic areas and with no requirements of high bandwidth Agriculture/Aquaculture productivity and safety monitoring Saltwater intrusion impact monitoring Wastewater monitoring Smart city monitoring
- 6. Data/analytics requirements and issues Farmers want to buy data and might provide/share devices but only related to their own local space Sharing data for local view among farmers/local agencies and for global view needed by government/agency Cloud is expensive when there is no need for farmers/local view LoRAWAN network-centric designs: Centralized analytics is not suitable IoT data-as-a-service and IoT infrastructure-as-a-service are not well support with multiple stakeholders Global IoT Summit 2018, 4 April 2018, Bilbao 6 Picture from Tran-Vu Pham for shrimp/fish/lobster farms monitoring
- 7. Why do we augment LoRaWAN? Support IoT service models: data-as-a-service model and infrastructure-as-a- service at the edge Support edge analytics with IoT data Work along with current LoRaWAN features for clouds: edge + cloud Focus on data analytics workflows as the main LoRaWAN applications We concentrate on software architecture design and configuration for edge analytics Global IoT Summit 2018, 4 April 2018, Bilbao 7
- 8. IoTRACE Framework Global IoT Summit 2018, 4 April 2018, Bilbao 8 Key design: making data available in the edge requires many changes in data flows and stakeholders management Enable edge computing
- 9. Stakeholders management for IoT data services Model stakeholders to support IoT data-as-a- service + IoT infrastructure-as-a-service Devices + data contracts Enable extraction of data at the edge for edge computing Global IoT Summit 2018, 4 April 2018, Bilbao 9
- 10. Network Server enabling edge computing Forward message to edge analytics servers Extract application data at the edge Global IoT Summit 2018, 4 April 2018, Bilbao 10
- 11. Extracting data in the edge Global IoT Summit 2018, 4 April 2018, Bilbao 11 Limitation mixed model: only subscription of devices
- 12. Current prototype Testing purposes: emulated devices/sensors Using LoRaWAN python library (https://github.com/jeroennijhof/LoRaWAN )to create LoRa messages Send to gateway via queue or to Packet Forwarder using LoRa-Gateway-Bridge (https://www.loraserver.io/lora-gateway- bridge/overview/) Emulating sensors read real dataset or emulate real sensors Global IoT Summit 2018, 4 April 2018, Bilbao 12
- 13. Current prototype Gateway and Network Server using python and MQTT Only for proof-of-concept of the features we augment Subscribers and data contracts managed by MongoDB Edge analytics components and functions: Python-based libraries and queues Javascript is being developed Global IoT Summit 2018, 4 April 2018, Bilbao 13
- 14. Simple example of analytics Global IoT Summit 2018, 4 April 2018, Bilbao 14 Analyti cs IoTRACE python libraries for edge computing deploy
- 15. Node-RED flows accessing edge data queues Edge analytics is written in Node-RED Application user deploys adaptor to the edge server working with Network Server Global IoT Summit 2018, 4 April 2018, Bilbao 15
- 16. Cross edge cloud analytics example Global IoT Summit 2018, 4 April 2018, Bilbao 16 Example Apex code We still need to mange two sides separately The queue identification for linking the edge to cloud is application-specific Cloud analytics
- 17. Conclusions and future work We enable diverse types of IoT service models and different configurations for LoRaWAN Augment software architectures for enabling IoT data- as-a-service and infrastructure-as-a-service at the edge Provide mechanisms for enabling edge computing for IoT applications Future work Refinement of architectures and data contract models Data interoperability for edge analytics for network of Network Servers Serverless functions for analytics & performance tests Global IoT Summit 2018, 4 April 2018, Bilbao 17
- 18. Thanks for your attention! Hong-Linh Truong Faculty of Informatics TU Wien, Austria rdsea.github.io Global IoT Summit 2018, 4 April 2018, Bilbao 18