Iot conference-oct-2014
2 likes1,349 views
The document discusses the Internet of Things (IoT) and its differences from the traditional internet, highlighting that IoT will comprise significantly more devices that are simpler, focused on machine-to-machine interactions, and often resource-constrained. It introduces the 'chirp' protocol, designed to handle the unique challenges of IoT by enabling lightweight communication with minimal overhead while allowing for self-classification and publish/subscribe dynamics for data flows. Lastly, the document emphasizes the need for open-source standards and incremental security measures to effectively manage the large-scale data generated by these devices.
Iot conference-oct-2014
- 1. Rethinking the IoT … with “Chirps” Francis da Costa Presented IOT Conference Oct 2014
- 2. The Internet of Things: It’s Different Out There The IoT won’t be much like the traditional Internet: • Scope – it’s 100 times bigger • Simple – majority of end devices “dumb” • Scalable – All control cannot be centralized • Subscription-based – too much data otherwise • Standards – primarily open-source with extensions • Security – must be incremental www.chirp-networks.com 2
- 3. Internet Core Scope: 100+ Times Larger Traditional Internet • 3 billion to 5 billion devices • Generally “always” On • Multi-purpose computers • Mostly for human interaction • Not resource-constrained Internet “Edge” Internet of Things • 700 billion to 1 trillion devices • Temporal: “sometimes” on • Specific-purpose machines • Mostly M2M interactions • Many resource-constrained www.chirp-networks.com 3
- 4. The Challenge of Truly Simple Devices • Up to a trillion of them • Sensors, actuators, appliances • Single-function, extremely limited “vocabulary” • Different in nearly every way from traditional Internet www.chirp-networks.com 4
- 5. Simpler Devices Will Rule the IoT Next wave of the Internet is Machines-to-Machines Ecosystems Human Oriented Ecosystem Machine Oriented Ecosystem Not (really) Resource Constrained • Lots more Processor, Memory, Protocol stacks • Human Oriented Consumption (external) • Assumed often “Always On” • Centrally-managed naming (MACID, et al) Often Resource Constrained • Often Limited or no processor, memory, etc. • Consumption for local use (Internal) • Many remote with Intermittent power • Built by millions of manufacturers worldwide …many cannot afford traditional IP protocol overhead www.chirp-networks.com 5
- 6. IoT Data Characteristics Machine to Machine (M2M) • Terse – not oriented to humans • Repetitive • Individual messages not critical • Meaning comes from combination with other data sources – “Small Data” • Consumption and generation is mostly Local • Usually unidirectional • Self-classified: new and necessary concept www.chirp-networks.com 6
- 7. Scalable Communication Lesson from Nature: Pollen Pollen propagates everywhere, but only specific receivers decode “message” www.chirp-networks.com 7
- 8. The “Lightness” and Elegance of Pollen • Self-classified (by species) • Extremely lightweight • No inherent transport mechanism • Uni-directional • Single-function • Individual “message” not critical • Receiver-oriented sensitivity … these are the reasons that pollen scales www.chirp-networks.com 8
- 9. The “Heaviness” of IPv6 (for Simple Devices) • Protocol overhead • Full error-detection/ retransmission • Bi-directional peer-to-peer • Multi-function: not for IoT • Transmitter-oriented networking … unneeded for the vast majority of the emerging IoT www.chirp-networks.com 9
- 10. IoT Data is Different, Protocol Must be Different: Chirps IPV6 Overhead = 40 bytes, 1 byte payload Chirp Overhead = 4 bytes, 1 byte payload Note that Chirp lacks: • Universally unique ID • Error correction/retransmission Minimal overhead for end device, but it must be applied elsewhere: -“Propagator” Nodes/Network, described later. Public Chirp Classification Private payload www.chirp-networks.com 10
- 11. Self Classification Lesson from Nature: Birdsong All birds derive some information, but only specific receivers fully participate www.chirp-networks.com 11
- 12. Self-Classification the Key to Discover / Subscribe • Massive amounts of data published from trillions of devices • Servers (“Integrator Functions”) will discover and subscribe to interesting small data flows • This requires self-classification at end device - Basic: Type of device (moisture sensor, streetlight, etc.) from open-source taxonomy - Incremental: unlimited additional classification through private fields • Published data may be open to all or proprietary www.chirp-networks.com 12
- 13. Self-Classification Lesson from Nature: Affinities Underlying event not seen, but affinities are visible www.chirp-networks.com 13
- 14. Known Publish/Subscribe Affinity (“Pollen”) Smoke Alarm Ventilation Lawn Sprinkler HHoommee NNeettwwoorrkk Time of Day Utility Pricing Local Weather Forecasts Home “Health” Advisor Application subscribes to many data streams www.chirp-networks.com 14
- 15. Discovered Publish/Subscribe Affinity (“Pheromones”) Affinity by time-of-day correlation: elevator activity Initial application: air conditioning control Affinity by location: lighting control Affinity by type of data: peak energy cost variations Integrator function seeks additional candidate data sources by affinities. Builds more refined causal models. Accelerate Learning through Affinities www.chirp-networks.com 15
- 16. Scalable Publish / Subscribe for the Edge • Discovery: Many useful data sources may be unknown. - Self-classification (“motion sensor”, “irrigation valve”, etc.) permits discovery of data “affinities” - Open Source top-level taxonomy crucial to scale and discovery • Subscriber-Based: Small data flows may be discovered, selected, and incorporated by Integrator Functions • Dynamic: New flows may be added and existing sources aged-out over time • Real Time: Reports, alarms, trends provided to humans via Integrator Function www.chirp-networks.com 16
- 17. Topology Lesson from Nature: Trees End devices don’t communicate with one another, so “tree” better than “web” www.chirp-networks.com 17
- 18. Emerging Tree Based IoT Architecture End Devices Propagator Mesh Network Filter Gateway Integrator Function Chirp Data Streams “Small” Data Flows “Big Data” Analysis www.chirp-networks.com 18
- 19. Scalability of IoT Architecture • End Devices can be cheap, simple, low power, unmanaged • Protocol sophistication only in Propagator Nodes • Prune and trim broadcasts – building “buses” • Optional distributed intelligent agents in Propagator Nodes - Extend subscription preferences of Integrator Functions - Add security and proprietary functions - Extends “Software Defined Network” publish/ subscribe functionality to edge of network www.chirp-networks.com 19
- 20. Scalability: Broadcasts, Distribution, Addressing Chirps from end devices Collection GPS Location Pruning Bundling IPv6 Packet Propagator Node Context Temperature Motion Propagator Nodes create “Small Data” flows from Chirp data streams www.chirp-networks.com 20
- 21. Scalability: Loading “Buses” Chirps to-and-from end devices Buses to/from different destinations Chirps unloaded/ reloaded Propagator Node Buses to/from different integrator functions Real Time Scheduling at Propagator Node www.chirp-networks.com 21
- 22. Propagator Node – Chirp-Ready IoT “Hub” IPv6 Serial Wired or Wireless Chirp, IP, or Legacy Bluetooth Power line Infrared WiFi www.chirp-networks.com 22
- 23. Propagator Node Form-Factors Infrared LED transceivers Power Replication USB, Ethernet Etc. Propagator node integrated with Smart Meter AC Mains faceplate replacement Propagator node combined with Wi‐Fi Access Point www.chirp-networks.com 23
- 24. Propagator Nodes –Networking Capabilities • Developed on Open Source platform: OpenWrt, et al • Build structured trees among themselves - Path discovery, routing, redundancy, fail-over - Simplicity through “near-optimal” routing • Manage multicast: pruning, forwarding, spoofing, etc. • Optional integrated Publishing Agents participate in publish/subscribe bus, machine learning • Offer wide variety of end device interfaces: wired, wireless, optical, etc. www.chirp-networks.com 24
- 25. Current Development Focus • Extending patent portfolio, existing code base • OEM licensees and Military projects underway • Now developing Propagator Node and Chirp-to- IP code for OEM customer applications • Offshore team in India for custom development www.chirp-networks.com 25
- 26. Chirp Protocols for the Emerging IoT • Simpler protocol for simple devices • Self-classified for scalability • Permits publish / discover / subscribe • Standards – open-source with extensions • Public and private fields for extensibility • Control loop flexibility • The challenge is scale, not address space www.chirp-networks.com 26
- 28. M2M Lesson from Nature: Autonomy and Control Devices operate independently, but may act in concert with external “cues” www.chirp-networks.com 28
- 29. Lower-Level Autonomy, Higher-Level Oversight IP IP IP IP Round-Trip Control Loops IP “Propagator” Lightweight chirp protocol only IP IP Dual Isochronous Control Loops IP overhead for every end device Traditional Emerging IoT www.chirp-networks.com 29
- 30. Progressive Refinements in Machine Control Loops Local Low Level Control L Lawn Sprinkler Actuator Valve Remote Control Model refinement Integrator Q: Did Sprinkler Turn On? A: No. R Propagator Relay Node Local Integrator CAUSAL INFERENCE ENGINE 1. Upstream Water Pressure good 2. Downstream Flow Absent Real and Virtual Sensor Publish/Subscribe Stream “ON” “OFF” Upstream Remote Pressure Sensor Down Stream Remote Flow Sensor S www.chirp-networks.com 30
- 31. Self-Classification (Pollen-like) for IoT: “Chirps” Public Section (mandatory) Private Section (optional) Marker Pointer (4 bits) 8 bit Marker 4 bit Marker 12 22 243 16 23 255 4 251 6 Agent ID Agent ID Total chirp length with 2 Byte Public Field, 4 bit Marker, 1 Byte Payload = 5.0 bytes 5.0 Bytes with 1 Byte Payload 6.0 Bytes with 2 Byte Payload 7.0 Bytes with 3 Byte Payload 04 22 243 06 02 255 03 8.0 Bytes with 4 Byte Payload www.chirp-networks.com 31
- 32. Importance of Open Source for Chirp Protocols • Too big, too much data, too unmanageable • Lessons from nature - Only publish /discover / subscribe can scale - Self-classified data necessary so receivers may select • Open Source taxonomy is scalable and distributed. • Top Level Taxonomy is open. • Extensible Sensor type Library • Extensible Actuator type Library • Enterprises and OEMs may develop custom and proprietary extensions for Private fields www.chirp-networks.com 32
- 33. Even “Closed” Flows Will Offer Open Source Information • Specifics of data stream may be private, but “affinities” are still observable - Information about number, location, and activity of devices (and much more) - Adds information to Open Source Taxonomy • Analogous to Twitter Discovery Functions • Every additional end device potentially adds to Open Source knowledge base www.chirp-networks.com 33
- 34. Security Must be Incremental to Open Source Format • Basic Chirp published and open to all - As in nature’s pheromones and birdsong • Private fields within Chirp may create “lock-and-key” relationship in OEM and proprietary applications - As in pollen – receiver determines • Further security achieved through distributed agents in Propagator Nodes • Secure data may still flow through Propagator Node network with open data, but is unintelligible - From nature: air transports both proprietary (e.g., pollen) and open “signals” (e.g., pheromones) www.chirp-networks.com 34
- 35. Security Must be Incremental DNA Pointer: 4 bytes, 8 bit Marker (1010) 12 22 243 16 23 255 4 251 6 Public Section (mandatory) Private Section (optional) Public Agent ID is 4.8.255 (4 byte Public, 8 bit Marker, DNA 255 (Subscribed) Agent states: Classification is 8.8.8.8 (1 byte each) Decrypted Chirp Class: 4.8.22.243.16.23. Its payload requires another Agent Private Agent 1.4.6 (for 4.8.22.243.16.23) decodes value (251) Chirp with public (open) payloads have shorter classifications e.g. Chirp Class 4.8.22: Temp=243F Pressure=16psi Humidity=23%. Larger Packets intended for slower transport. Enterprises may define their (internal) classification schemes. Discovery of “unknown” chirp classes detected, addressed in SIGs. Distributed, organic growth of chirp classification taxonomy. www.chirp-networks.com 35