Virtualize IoT Multi-RAT Stacks at Edge A work in collaboration with RISE SICS Presentation for EUCNC18 Workshop 3: Multi-provider, multi-vendor, multi-player orchestration: from distributed cloud to edge and fog environments in 5G Chenguang Lu, Daniel Cederholm and Maximillian Stiefel Ericsson Research 2018/06/18 Ericsson Internal | 2018-02-21
I o T challenges Business and operation challenges Data: what Customer Business Internal silos and where? outcomes objectives Technology Multiple Security and Interoperability fragmentation vendors privacy Ongoing Future Scalability Integration measurement proof Most of them are about technologies! Ericsson Internal | 2018-02-21
Fragmentation over protocol stacks Cloud platform approach Application domain: vertical and horizontal, APP semantics (IPSO, Zigbee Alliance, OMA LWM2M, UPnP Forum etc.) Platform CoAP , MQTT, HTTP … Transport Current GW approach Network (IP) IPv4, IPv6, non-IP proprietary Wireless Mesh Transceiver domain MAC More than 40 existing protocols (RATs) More fragmented here! e.g. Zigbee, BLE, WiFi, NB-IoT, LoRa etc. PHY — To support multi-RATs, parallel IoT networks need to be deployed — Drives up costs — Not scalable Ericsson Internal | 2018-02-21
Virtualize IoT multi-RAT at Edge Cloud -Cloud-RAN approach for IoT — Reference approaches: single RAT — WiFi AP-AC approach: real time function in AP and non-real time function in AC. — Cellular C-RAN/V-RAN approach: centralize processing for performance, cloudify high layers for scalability and cost reduction — Extend such approaches for Multi-RAT IoT: IoT Cloud RAN at Edge — Radio infrastructure: IoT radio heads handling multiple IoT radios, e.g. BLE, Zigbee, NB-IoT, LoRa etc. — Edge Cloud: virtualized full-stack IoT multi-RAT functions, i.e. PHY, MAC, NET …, in an Edge Cloud environment Ericsson Internal | 2018-02-21
IoT challenges addressed Data: what Customer Business Internal silos and where? outcomes objectives Technology Multiple Security and Interoperability fragmentation vendors privacy Ongoing Future Scalability Integration measurement proof Softwarizition of lower layers helps address these challenges Ericsson Internal | 2018-02-21
PoCdesign example: IEEE 802.15.4 CoAP + DTLS + HTTP MQTT .. CoAP .. LWM2M + IPSO objects — Full stack implementation TCP UDP — PHY: GNU Radio IPv6 (+ 6LoWPAN) + RPL — MAC and higher: Contiki-NG Wifi 802.15.4(e) PLC Bluetooth .. — Early integration: UDP socket Contiki-NG — Virtualization: Docker POSIX socket — Radio head: SDR-based — FH interface: IQ samples over Ethernet GNU Radio Edge (15.4) (currently ZeroMQ) — Challenges: Ethernet — Decrease latency USRP/ — Decrease jitter Radio head LimeSDR Ericsson Internal | 2018-02-21
Testbed design: 15.4 + NB-IoT Edge Radio head 2.4GHz, 15.4 USB USRP Zolertia Firefly 15.4 Ethernet PC/ PC switch Server USB USRP USRP Sub GHz, NB-IoT Virtualized 15.4 NB-IoT UE and NB-IoT stacks (SDR-based) — EUCNC demo: first functional-level software implementation and system integration, 15.4 full stack and NB-IoT DL (key PHY functions) dockerized at edge — Next step: more measurements and results analysis, and then further improve the implementations, e.g. in radio performance, reliability, latency etc. Ericsson Internal | 2018-02-21
See you in Booth 02!
Recommend
More recommend
