SLIDE 1
Designing Delay-Tolerant Data Services for the Network of Things - - PowerPoint PPT Presentation
Designing Delay-Tolerant Data Services for the Network of Things - - PowerPoint PPT Presentation
Designing Delay-Tolerant Data Services for the Network of Things Daniel Austin Interstellar Travel, Inc. 1st Annual Big Data Innovation Summit daniel@thestarsmydestination.com April 09 2014 Silicon Valley, California Big Ideas for Todays
SLIDE 2
SLIDE 3
“If we had computers that knew everything there was to know about things – using data they gathered without any help from us – we would be able to track and count everything, and greatly reduce waste, loss and cost. We would know when things needed replacing, repairing or recalling…”
- Kevin Ashton, 1989
The Network of Things
SLIDE 4
A Day in the Connected Life
…”a system where the Internet is connected to the physical world via ubiquitous sensors…”
SLIDE 5
Characteristics of the Network of Things
- Many small messages
- Intermittent transmission
- Connectionless
- Stateless
- No guaranteed delivery
- Lazy/No Acknowledgements
- Mesh architectures based on proximity
- Mixed/variable security
SLIDE 6
Service Types for the Network of Things
- Proximity
– Identity – Authorization/Eligibility – History – Personalization
- Location
- Companions/Presence
- Service Discovery
- Ordering/Billing/Payments
SLIDE 7
WPAN (NoT) Protocols
- IEEE 802.15.4 was designed for wireless PANS
– ZigBee and other protocols
- 6LowPAN (RFC 6282) describes how to route IPv6 over
802.15.4 networks
- Problems with IPv6 & NoT
– IPv6 requires minimum MTU of 1280 bytes
– Far too large for IoT messages – Overhead for addressing ~ 40 bytes of IPv6 + 20 bytes for TCP
– IP designed for bulk data transport
– Congestion is not an issue for the NoT!
SLIDE 8
Bluetooth Low Energy vs 6LowPAN
BLE
- Range = 100m
- Security = 128-bit AES
- MSG SZ = 128 octets
- BLE 4.0 protocol
- Clean separation of BLE protocol stack from TCP/IP
6LowPAN
- Range = 20m
- Security = 128-bit AES
- MSG SZ = 127 octets
- IPv6 over 802.15
- Mixed protocol stack
SLIDE 9
Delay-Tolerant Networking
- TCP/IP Assumptions:
– End-to-End connection – Short, fixed delays – Symmetric data rates – Low error rates – Some knowledge of existing network
- DTNs originated at NASA for interplanetary communications (RFC 4838 & 5050)
– Applies to all intermittently connected scenarios, including the NoT
SLIDE 10
How DTNs Work
TCP/IP DTN
Source; DTN SIG
SLIDE 11
DTNs: From Cars to Interplanetary Networks
Source; DTN SIG
SLIDE 12
CAP Theorem & DTNs
Source: DTN SIG
SLIDE 13
Big Data and the Network of Things
- Many Small Devices = Big Data
- Consistency, Availability, and (network) Partition take on
new meanings in the NoT
– DTNs weaken CAP assumptions – Consistency can’t easily be checked
- Big Data, NoT, and Security
– Not based on encryption – Anonymity through disaggregation
SLIDE 14
Are You a Big Data Problem?
- Each person will generate roughly 20 petabytes of data over
the course of a lifetime
– Users have little control over collection and storage – Email, documents, receipts, bills (!), your car, music, books…
- Pervasive computing multiplies the problem
– Majority of data valueless out of context – Security & Privacy concerns
- Networks of Things instead of ‘Internet of Things’
– Security by data partition
SLIDE 15
Ubiquitous/Anonymous Peering Patterns
- Peer-to-peer
- Low levels of security
- Resilient to network
partitions
- Proximity based – no
identity
- Good for service
discovery
SLIDE 16
Client/Server Patterns
- Clients transmits to one
server
- Higher Security
- Requires server asymmetry
- Can offer additional services
– History – Personalization – Identity
SLIDE 17
Design Rules for Data Services
- Security & Privacy First!!
- Delay tolerance required
- Idempotent/stateless
- Messages vs. Request/Response
- Anticipate Maximum Mesh
- Batched/Bundled vs. Event-Driven
SLIDE 18
Big Takeaways
- IoT <-> DTNs <-> Big Data
– pervasive = invasive ? – Evolution of Big Data Depends on the IoT
- Delay-tolerance for mobile networks
– Realistic assumptions – CAP theorem, store-and-forward consistency – Online/offline distinction is blurring
- Only expect partial adoption, based on loose aggregations
– ‘Networks of Things’ vs. ‘Internet of Things’
SLIDE 19
By 2020 everyone, everything and everywhere will be connected in real time. More than 50 billion connected devices will be used in the Networked Society.
Source: http://www.ericsson.com/thinkingahead/networked_society
SLIDE 20