![IEEE SmartTech 2018
( http://www.ieeesmarttech.com/ )
Future of
Edge Computing
in Mobility Systems
Narayanan Subramaniam [ Chief Software Architect @ ]
LinkedIn: http://www.linkedin.com/in/cnsubramaniam
@copyright: Narayanan Subramaniam](https://image.slidesharecdn.com/future-of-edgecomputing-in-mobility-180826034933/85/Future-of-Edge-Computing-in-Mobility-Systems-1-320.jpg)
Future of Edge Computing in Mobility Systems
- 1. IEEE SmartTech 2018 ( http://www.ieeesmarttech.com/ ) Future of Edge Computing in Mobility Systems Narayanan Subramaniam [ Chief Software Architect @ ] LinkedIn: http://www.linkedin.com/in/cnsubramaniam @copyright: Narayanan Subramaniam
- 2. Context: (F)Edge Computing @ Smart Battery Swap/Charge Stations) SMART Network IoT Hub, Monitoring, Provisioning, Machine Learning Legend: Metrics, Telemetry Data Alerts, Audits Software (Features, Algorithms – DevOps) Diagnostics (Logs, Events) Configuration, Provisioning, Billing • DevOps based Design with Agile Development • Security, Scale, Availability and Performance is Implicit • Intelligent Autonomous Stations and Smart Batteries Swap/Charge Stations) @copyright: Narayanan Subramaniam
- 3. Landscape of Computing Use Cases for Mobility Consumers and Fleet Operators • Personalization: • User Experience • Search • Performance • Utilization and ROI • Collaboration • Meetings • Content • Augmented Reality • Search • Weather • Points of Interest Vehicles and Mobility Stations (e.g. Swap, Charge) • B2B: • Asset Tracking, • Operational Metrics • Resource Optimization • B2C • User Experience • Performance • Transaction Metrics • Autonomous Vehicles • Safety and Security • Availability Energy Providers (e.g. Renewable Sources) • B2B: • Dynamic Pricing • Energy Demand • Grid Feedback CONTINUOUS DATA STREAMS AT VOLUME, IN VARIETY, THAT ARE EVER CHANGING 1. Who: Sensors, Embedded Computing, Personalization 2. When: (Near) Real Time, Accumulated Data 3. Why: Operations, Performance, Safety, Reliability, Availability and Data Insights (AI/ML) @copyright: Narayanan Subramaniam
- 4. Why Embark on (F)Edge Computing for Mobility Solutions ? @copyright: Narayanan Subramaniam Near Real Time Processing, Autonomous Operations vs Cloud Latency/Availability • Operations: • Multiple streams of sensor and embedded based data e.g. Power Delivery, Charging, Braking • Thermal, Energy, Policy Management • Metrics Aggregation and localized Decision Algorithms, A/B Testing • Safety and Security • Anomaly Detection and Isolation • Fire Detection and Suppression • Video-analytics – theft, environment • Availability • Fault Patterns and Redundancy • Security Denial of Service issues • Personalization • Secure content caching, offline modes for Billing/Payment Cost of Computing-In and Connectivity-To the Cloud, Vendor Lock-In • 3-5 Year Cost Analysis: • Storage and Archival • Connectivity and Data Transit (APIs) • Computing (Serverless and Server Based) with ever increasing Data • Technology Lock-In • Lock-In to vendor extensions • Data extraction and portability costs • Delicate Balance • Skills vs Recurring Non Core Cost • How critical is Autonomy ? Privacy, Data Retention, Regulations/Compliance • Data Retention: • Locally Managed – push only what is needed in the Cloud • Privacy • Cloud Data Storage Privacy versus local Aggregation and/or Anonymization/Pseudonymisation of PII Data • Explicit Opt-In/Out Costs • Data Sovereignty • Geo-location of Data At Rest
- 5. Reference Architecture for (F)Edge Computing Credits: Reference from OpenFog Consortium: https://www.openfogconsortium.org/wp- content/uploads/OpenFog_Reference_Architecture_2_09_17-FINAL.pdf
- 6. Protocol Adaptation, Model Adaptation Device Addressing and Routing High Availability Messaging Wireless (Bluetooth, Wifi, GPRS, GPS, NFC) Wired (CAN2.0, Modbus, Ethernet, TCP/IP, USB, GPIO, SPI) Cloud/IoT Interfaces (MQTT, REST) Database: • Metrics • Alarm/Alert • Audit, Transactions • Event/Logs • Configuration • Toggle, A/B • Diagnostics • Security • Algorithms/Policies Station Sensor and Battery Pack Interfaces Station Application Stack: • SWAP and Charge Orchestration • Station Monitoring • Energy Management • Safety and Security Management • Charging and Thermal Management • Diagnostics • Configuration and Provisioning • SW/FW Management • Entitlement and Billing • UI/UX • Containers • Localized ML, Analytics High Level (F)Edge Computing Architecture @ @copyright: Narayanan Subramaniam
- 7. Mobility (F)Edge Computing Technology Stack Candidates Credits: Images Thanks to Open Source Community Technology and Development Organizations
- 8. Challenges and The Way Forward @copyright: Narayanan Subramaniam Skills and Cost • Skills: • Multidisciplinary – Agile Software Engineering, DevOps, PC, Storage technologies, Networking/Telecom, Cloud, Machine Learning experience • Full Stack, Embedded Developers, need retraining in orthogonal skills • Cost: • Buy vs Build results in vendor lock-in versus skills challenges • Multidisciplinary skills to use or build the (F)Edge Platform effectively in a Buy vs. Build approach • Cost Effective Industrial Grade components for a range of Thermal, Environmental, Serviceability needs Communication Interfaces for (Big) Data • Data Is Key: • Protocols like CAN and Modbus not appropriate for Bulk Data Transfer • Lower Cost Industrial Ethernet/Wifi and Convertor components to facilitate exploding Data requirements • Clear separation of Control, Management and Data Plane with Enterprise grade Security Domain Specific Data Standardization • Standardized Data Models that are Domain specific for common API level Orchestration and off the shelf Software components to solve baseline Use Cases • Data Model Extension constructs for business differentiation and competitive benefits • Design for Privacy
- 10. THANK YOU ! @copyright: Narayanan Subramaniam