DCIT 2015 Wuhan, Hubei, China, November LIMOS UMR 6158 CNRS, Clermont-Ferrand, FRANCE

Transcription

1 DCIT 2015 Wuhan, Hubei, China, November 2015 LIMOS UMR 6158 CNRS, Clermont-Ferrand, FRANCE 1

2 Outline Introduction: Panorama of IoT technology IoT core technologies: IoT node hardware: State-of-the-art of IoT Node Trend and challenges - Communication protocol stack: State-of-the-art: 6LoWPAN and RPL Challenges Middleware: CoAP, OASIS /MQTT State-of-the-art CoAP (IETF) Trend and challenges Conclusion Open research issues 2

3 Introduction IoT Internet of Things and WoT Web of Things : Emergent and multidisciplinary science, very active and competitive research area. IoT and WoT have unlimited potential applications: air, underground and underwater. Source: 6LoWPAN: The Wireless Embedded Internet Companion Lecture Slides 3

4 At least 26 billion devices will be connected on the Internet by 2020 (Gartner), IoE Internet of every Things creates $19 trillion of value at stake for companies and industries (Cisco) IoT product and service suppliers will generate incremental revenue exceeding $300 billion in 2020 (Gartner). IoT will be the next IT revolution: Industry 4.0, IoE IoT is one of the key issues to support sustainable development (smarter planet smart every where : smart grid, smart home, smart building, smart care, smart farming ). 4

5 5 Panorama of IoT platforms Big ICT players: IBM (Bluemix), Microsoft (Azure), SAP (HANA) provide the IoT cloud-based platforms containing three mains layers: Back-end: Integration and Services Middleware: connect and collect. Front-end: IoT nodes The front-end layer is open for diverse players to develop their specific physical devices for specific application (NB- IoT, ISM (IEEE , Sigfox, LoRa ) Figure 2: IBM Bluemix IoT based platform

6 Y Key components of IoT core technology Radio Radius:100 Fieldsize X:500 Fieldsize Y:500 Node Number:50 Minimum Intersection Nodes:1 Optimum Intersection Nodes:2 IoT Cloud-based platform 500 N21(15) N33(15) B35(9 15) M9 M3 N8(3) N49(3) 400 M15 N34(15) B28(9 15) B22(3 9 10) 300 B12(15 26) N13(26) M26 B27(9 10) M10 M2 N32(10) N42(2) B29(2 10) N6(10) B48(2 5) B45(2 5) M5 N38(5) B11(14 26) 200 B19(14 26) 100 N46(14) M14 B39(14 18) B31(18 41) N43(18) M18 N37(41) N23(41) N40(41) M41 B16(18 41) N36(41) N24(7) B4(1 N20(1) 47) M47 B17(1 47) N30(1) N44(1) M1 N50(1) Simulator 0 N25(18) M7 Master: 13(26%) Lost: 0(0%) Slave: 37(74%) Bridge: 15(30%) Slave without Intersection: 22(44%) X Applications Middleware Protocol Stack Operating system WSN Node HW = 6

7 IoT core technology: IoT node - Basic Hardware Key features of a IoT Node Sensor Signal conditioner Processor Memory: RAM & ROM Peripheral Devices: ADC, UART, SPI, I²C, GPIO Power supply & Power management Unit Microcontroller Wireless Access medium Antenna 7

8 IoT Node HW: State-of-the-art and design trend Two main development and design trends: Commercial Off-The-Shelf COTS and System on Chip SoC COTS: platform for test and validation, real world experimentation SoC: Ultimate goal to achieve the implementation of long lifetime, low cost and invisible IoT node integrated and embedded into environment or object. 8

9 System On Chip: SoC 9

10 IoT node HW: trends and challenges Trends and challenges: More powerful IoT node: 32-bit core (e.g. WMSN) Robust and Energy Efficient Multicore (lifetime): Energy efficient IEEE : ~45mA ~15mA (2Mbs) (research prototype <10mA) Texas Instruments MSP430 runs at less than 100µA/MHz (FRAM ferroelectric read-only memory) Radio cognitive: IEEE , NB-IoT Battery-less IoT node: Energy harvesting. 10

11 IoT core technology: Operating System State-of-the-art: Event-driven: TinyOS, Contiki, MUROS Multi-thread: RETOS, tkernel, ScmRTOS (RTOS), MANTIS, AmbientRT etc. Hybrid: Contiki (not native), HEROS (native). Event-driven is not adapted to: Complex hard real-time application Distributed Component Based Model implementation (collaborative processing ) Multi-connection server (CoAP: delay) 11

12 IoT core technology - Operating System: Challenges Challenges: Energy Efficient (resource-aware: multicore and robustness), real-time collaborative processing and small memory footprint. Operating system is the key technology to implement time redundancy fault tolerant IoT node. 12

13 IoT core technology: communication protocol stack Many dedicated routing protocols were developed: Robustness, scalability and interoperability are still an open question. IPv6 for IoT: 6LoWPAN&RPL (IETF) Name OS platform Standard Note version ContikiRPL Contiki RFC6550 Support officially TelosB platform OpenWSN OpenWSN RFC6550 TinyRPL TinyOS Draft-ietf-rollrpl-17 NanoQplus NanoQplus Draft-ietf-rollrpl-13 Support TelosB and Epic platforms (only storing mode and single DODAG) RPL implementations for real-world devices [Chen Yibo, PhD thesis] 13

14 Resume 6LoWPAN Two main implementations: ContikiRPL SICS: Swedish Institute of Computer Science TinyRPL UC Berkey Simulator: Cooja (SICS) Small resource: Low power CPU and memory (<30KB of memory) Challenges: Large scale network still needs to be tested and evaluated (i.e. DIO overhead) Edge router or LBR LLN Border Router may solve this problem (RPL). 14

15 State-of-the-art of Tiny Server (IETF): TinyCoAP, CoapBlip, CoAP for Contiki, libcoap, CoAPy, jcoap, Californium Memory footprint of CoAP for Contiki: 8.5KB of ROM 1.5KB of SRAM 15

16 Other protocols: WoT protocol standard? - MQTT (IBM), AMQP (NASA, Google), STOMP, RabbitMQ Item OASIS/MQTT CoAP Traffic flow P2PP, P2P, MP2MP P2P Communication protocol TCP UDP Computing model Client/Server Server License OPEN OPEN 16

17 WoT: trend and challenges Both OASIS/MQTT and CoAP will be the standard? Challenge: How to implement energy efficient mashups? Relevant data (uncertainty data), Volatile and persistent data (Big data) 17

18 Intermittent data loss (Lossy link, sleep&wakeup mode) Network dysfunction due to environment change. Nodes are always sleep: Network disconnected for unknown reason Nodes are always wakeup: Battery exhausted prematurely (try continuously to send message) Unknown reasons 18

19 Why IoT node fails? Inconsistent specification Software errors Message synchronization Logical (Operation) error Data consistency Hardware errors Transient faults Permanent faults Lack of resource (e.g. stack overflow) 19

20 Dependable IoT node Classical techniques: Primary backup, DMR and TMR enable to implement fault tolerant system (space and time redundancy). Formal languages How to implement low cost, fault tolerant and energy constraint IoT node? 20

21 Hardware failures High speed low power CMOS technology Hardware main sources of failure: More sensitive to external disturbances, such as radiation effects electromagnetic influence, parameter fluctuations, and high temperatures, aging, application dependent, latch-up etc. (Mixed-Signal IP and SoC Design ) Taking into account the current status of ultra low power CMOS and mixed signal technology, how to implement a low cost and robust energy efficient 21 IoT node?

22 Dependable IoT node: Open research issues Open research issues: auto fail detection and auto configuration Formal Software and auto-check Time redundancy: auto fail detection and auto configuration (combine software and ultra low energy hardware fail detector, multicore architecture) Soft errors recovery: restart, execute the substitute task Permanent errors: restart with new hardware configuration (multicore architecture) 22

23 Conclusion The weak point of IoT large scale deployment concerns its robustness at front-end level. How to design a low cost, fault tolerant and energy efficient IoT is an open research issue. The IoT will drive new research fields and uncountable applications. The IoT will revolutionize (big bang) the ICT and continue to push ahead the current trend: super data centers and smart tiny data centers (trillion?) in order to meet the requirements divers applications. The economic and social impact of IoT is an open question, but one thing is sure IoT will change the way of our every day living. 23

24 «This work was founded by the French National Research Agency, the European Commission (Feder funds) and the Région Auvergne in the Framework of the LabEx IMobS 3» 24