How new Low Power Wireless Area Networks (LPWAN's) are aggressively challenging the Internet of Things status quo and how industry can exploit this opportunity. Specifically, the ability to query IoT endpoints in real time, improve network capacity and data rates, and the ability to deploy a filesystem in order to create a "Hadoop"-like real-time query capability at the edge of the network is explored.

1. The iot
Hunger games
2015
ONE TECHNOLOGY IS DRIVING A
NEW WAVE OF INNOVATION FOR
THE INTERNET OF THINGS …

2. WHO WE ARE
We began driving innovations in the internet of things over 10 years
ago at our last company, Savi Technology and believe that the best way
to connect networks with many battery-powered sensors is not through
WiFi, Bluetooth, or cellular, but via something better.
We invented a better way of connecting things using very low power
and over long distances, a technology called DASH7. Our company also
builds tools, API’s, and software to make DASH7 more accessible to
developers.
We recently began receiving inquiries about a new class of IoT
modulation technologies called Low Power Wide Area Networks. We
think LPWAN’s are exciting and this presentation tells you why.

3. LOW POWER STATUS QUO
30 feet 3 miles300 feet
Short Range /
“WPAN”
Medium Range

4. NEW CHALLENGERS
Short Range /
Local Area
Medium Range
up to 30 Miles
Long Range /
“LPWAN”
30 feet 3 miles300 feet

5. RANGE IS MASSIVELY BETTER
up to 30 Miles
Long Range /
“LPWAN”
30 feet 3 miles300 feet
Short Range /
Local Area
Medium Range

6. 30 feet 300 feet 5 Kilometers
Long Range /
“LPWAN”
up to 30 Miles
FOR THE SAME PRICE
Short Range /
Local Area
Medium Range
Low Power Wide Area Networks
• Very long range
• Multi-year AA battery life
• Low cost: sub-$10 per node

7. THE FUTURE OF THE IOT
10 meters 5 Kilometers100 meters
Short Range /
Local Area
Medium Range
Long Range /
“LPWAN”
up to 50 Kilometers
We believe most wireless sensor networks will be LPWAN-
based, as LPWAN’s offer comparable pricing and power
consumption to legacy WPAN/WLAN options, but with:
• Significantly improved range and signal coverage
• Better monetization opportunities for customers

8. KEYS TO LPWAN LONG RANGE
+ +Sub-1GHz
Radio Bands
Really Low
Bit Rates
Frequency
Spreading
Longer wavelengths
allow vastly longer
range and lower
power consumption
Common bands
include 915, 868,
433, and 169 MHz.
Technology being
deployed in most
LPWAN modulation
schemes use some
form of spreading to
combat interference
Low data rates of just
a few hundred bps
increase range, but
as a result the
packets get very
“long”, which leads to
new challenges.

9. + +Sub-1GHz
Radio Bands
Really Low
Bit Rates
Frequency
Spreading
Longer wavelengths
allow vastly longer
range and lower
power consumption
Common bands
include 915, 868,
433, and 169 MHz.
Technology being
deployed in most
LPWAN modulation
schemes use some
form of spreading to
combat interference
Low data rates of just
a few hundred bps
increase range, but
as a result the
packets get very
long. This leads to
new challenges.
KEYS TO LPWAN LONG RANGE
These technologies for achieving long range are
old and well-established. Advances in
semiconductor technology over the last 40 years
are what enable low-cost and low-power.
So barriers-to-entry are also low …

10. COMPETITION ARRIVES …
Long Range /
“LPWAN”
up to 30 Miles30 feet 3 miles300 feet
Short Range /
Local Area
Medium Range

11. up to 50 Kilometers50 Kilometers
Long Range /
“LPWAN”
Long Range /
“LPWAN”
Medium Range
up to 30 Miles30 feet 3 miles300 feet
… AND INTEGRATORS
Short Range /
Local Area

12. 30 feet 3 miles300 feet
AND DOZENS OF STARTUPS
Short Range /
Local Area
Medium Range

13. so the games begin …

14. BUT DEVELOPERS HESITATE
1. Choosing wisely among multiple LPWAN suppliers,
including some which may disappear in a year or two,
is difficult.
2. There is no LPWAN PHY standard.
In fact, the three prominent PHYs are radically different.
3. No standardized networking stack.
4. Market is dominated by high cost, single-vendor silicon.
5. Scalability of some LPWAN technologies

15. MOST LPWAN TECH IS PHYSICAL
LAYER ONLY
15
OSI Layer
7 Application Undefined Undefined Undefined Undefined
6 Presentation Undefined Undefined Undefined Undefined
5 Session Undefined Undefined Undefined Undefined
4 Transport Undefined Undefined Undefined Undefined
3 Network Undefined Undefined Undefined Undefined
2 Data Link Partial Definition Undefined Partial Definition Undefined
1
Physical
“PHY”
LoRa @
169 - 960 MHz
Various @
315 - 930 MHz
SigFox @ 900, 868
MHz
SigFox and
Generic PHYs
Example LPWAN PHY’s

16. MOST LPWAN TECH IS PHYSICAL
LAYER ONLY
16
OSI Layer
7 Application Undefined Undefined Undefined Undefined
6 Presentation Undefined Undefined Undefined Undefined
5 Session Undefined Undefined Undefined Undefined
4 Transport Undefined Undefined Undefined Undefined
3 Network Undefined Undefined Undefined Undefined
2 Data Link Partial Definition Undefined Partial Definition Undefined
1
Physical
“PHY”
LoRa @
169 - 960 MHz
Various @
315 - 930 MHz
SigFox @ 900, 868
MHz
SigFox and
Generic PHYs
Example LPWAN PHY’s
The physical layer defines the way bits are
converted into radio signals: encoding, signal
modulation, the radio frequency to use, and
related low-level parameters.

17. Partial Definition Partial Definition
1
Physical
“PHY”
LoRa @
169 - 960 MHz
Various @
315 - 930 MHz
SigFox @ 900, 868
MHz
SigFox and
Generic PHYs
YET CUSTOMERS NEED MORE
THAN JUST PHYSICAL LAYER
• Addressing Options
• Networking Options
• Session Options
• Device Wakeup
• Authentication
• Encryption
• Device Filesystem
• Power Management
• Location-based Services
• Sensor Options
• Application API’s
• Device Management
UNDEFINED IN PHYSICAL LAYER

18. 18
OSI Layer
7 Application
Undefined Undefined Undefined Undefined
6 Presentation
5 Session
4 Transport
3 Network
2 Data Link Partial Definition Partial Definition
1
Physical
“PHY”
LoRa @
169 - 960 MHz
Various @
315 - 930 MHz
SigFox @ 868, 915
MHz
SigFox and
Generic PHYs
Example LPWAN PHY’s
HISTORIC OPPORTUNITY

19. HISTORIC OPPORTUNITY
19
OSI Layer
7 Application
Undefined Undefined Undefined Undefined
6 Presentation
5 Session
4 Transport
3 Network
2 Data Link Partial Definition Partial Definition
1
Physical
“PHY”
LoRa @
169 - 960 MHz
Various @
315 - 930 MHz
SigFox @ 868, 915
MHz
SigFox and
Generic PHYs
Example LPWAN PHY’s
Standardizing layers 2-6 will
accelerate LPWAN adoption
worldwide and basically make
many people happy.

20. THIS IDEA MAKES SENSE
1. Avoids fragmentation. Too many competing stacks over different PHY’s = slow growth.
2. Proprietary stacks are not portable across PHY’s. For example, SigFox’s stack only works
with SigFox’s own unique PHY and operating configuration. Similarly, stacks like
LoRaWan are limited to a single provider of silicon.
3. “Roll-your-own” stack inhibits developers and customers. A common stack gives
developers and customers the option to choose among PHY technologies and focus on
the application layer, while lowering maintenance and support costs.
4. Interoperability. Standardizing provides key elements of interoperability, creating new
product and application opportunities like multi-PHY gateways and endpoints, similar to
WiFi.
5. Performance improvements. Roll-your-own stacks will be slower to respond to
marketplace innovations as well as among PHY layer suppliers. A common stack makes
the trajectory of LPWAN’s more assured!

21. Requirement
Provide Robust
Networking Features
P2P, broadcast, multicast, and IP addressing. Ad-hoc networking. Rapid device
discovery. Deployable across global ISM bands, not just USA or EU. Improves
network capacity. Real-time locating system support.
Real-Time Data Collection
Some IoT technologies achieve long battery life using huge time intervals
between messages. Customers want their data when they want it and want to be
able to “Google” their network for a diverse range of criteria and data types.
Preserve or Improve
Long Range Messaging
Sounds obvious, but not all stacks can support the long range or cellular-like
design of LPWAN’s with a fully two-way system that does not compromise battery
life or network capacity.
Provide Maximum Practical
Security & Privacy
This is a big topic, but a LPWAN stack must at a minimum support a) MAC-layer
address encryption, b) AES, RSA, or ECC data encryption standards, and c) devices
must remain silent until awoken by an authorized device.
Preserve or Improve
Battery Life
It’s not enough to support long-range messaging. A stack must have a neutral or
positive effect on battery life without compromising latency or range.
WHAT THIS STACK HAS TO DO
(At a minimum)

22. Requirement 6lowPAN LoRaWAN Actility Linklabs
Haystack/
DASH7
Provide Robust
Networking Features
Yes Some Some Some Yes
Real-Time Data Collection No No No No Yes
Preserve or Improve
Long Range Messaging
No Yes Yes Yes Yes
Provide Maximum Practical
Security & Privacy
Yes No No No Yes
Preserve or Improve
Battery Life
No No No Some Yes
For a more detailed comparison, click here.
HOW TODAY’S STACKS MEET
FUTURE LPWAN REQUIREMENTS

23. • Combination of low-power, long-
range, low-latency, high security,
universal interoperability, and IP-like
data model is unique to DASH7.
• Lower Layers provide low-power,
long-range, low-latency, high security.
• Filesystem & Session are “glue” that
provide universal interoperability. No
Application Profiles
• Works with any application protocol
that can ride on UDP, SCTP, or NDEF/
NFC (e.g. CoAP, MQTT, AllJoyn…
many others).
Lower Layers
Application Layer
Physical
Data Link
Networking (M2NP)
Transport (M2QP)
SessionModule
Standard Apps Custom Apps
ALP Framework
FilesystemModule(M2FS)
M2DEF
RF
UI (opt.)
BASIC ARCHITECTURE

24. some technical background on
three important
haystack / dash7
features
24

25. Error Correction
Technology
None
Reed Solomon
(RS Code)
Voyager Code Turbocode LDPC
Used By
SigFox, ZigBee,
6LoWPAN, etc.
LoRa,
Data Storage
Voyager 2,
Haystack/DASH7
3G Cellular 4G Cellular
Signal Gain
(10-6
BER)
None
4 dB
(250%)
8 dB
(630%)
9 dB
(794%)
9.5 dB
(891%)
Supports Variable
Length Packet
Yes Yes Yes No No
Underlying
Technology
None
Iterated
Base-32
RS Code
Concatenated
Viterbi Code
with Base-256
RS Code
Fully Recursive
Convolutional
Code
Low Density
Parity Check
(LDPC)
Introduction Date
1850’s
(Morse Code)
1960’s
(Data Storage)
1980’s
(NASA)
1990’s
(Cellular)
2000’s
(Cellular)
1. ERROR CORRECTION
25

26. Error Correction
Technology
None
Reed Solomon
(RS Code)
Voyager Code
Used By
SigFox, ZigBee,
6LoWPAN, etc.
LoRa,
Data Storage
Voyager 2,
Haystack/
DASH7
Signal Gain
(10-6
BER)
None
4 dB
(250%)
8 dB
(630%)
Supports Variable
Length Packet
Yes Yes Yes
Underlying
Technology
None
Iterated
Base-32
RS Code
Concatenated
Viterbi Code
with Base-256
RS Code
Introduction Date
1850’s
(Morse Code)
1960’s
(Data Storage)
1980’s
(NASA)
26
• Haystack developed the
Voyager Code on ARM
• All things being equal, a
message transmitted
using DASH7 arrives in
less than half the time of
a LoRaWan message, or
at worst 1/6th of a
SigFox message.
• Reduce power by
transmitting less.
• Increase capacity of cell
by transmitting less.
1. ERROR CORRECTION

27. 27
1. ERROR CORRECTION
• Haystack developed the
Voyager Code on ARM
• All things being equal, a
message transmitted
using DASH7 arrives in
less than half the time of
a LoRaWan message, or
at worst 1/6th of a
SigFox message.
• Reduce power by
transmitting less.
• Increase capacity of cell
by transmitting less.
Error Correction
Technology
None
Reed Solomon
(RS Code)
Voyager Code
Used By
SigFox, ZigBee,
6LoWPAN, etc.
LoRa,
Data Storage
Voyager 2,
Haystack/
DASH7
Signal Gain
(10-6
BER)
None
4 dB
(250%)
8 dB
(630%)
Supports Variable
Length Packet
Yes Yes Yes
Underlying
Technology
None
Iterated
Base-32
RS Code
Concatenated
Viterbi Code
with Base-256
RS Code
Introduction Date
1850’s
(Morse Code)
1960’s
(Data Storage)
1980’s
(NASA)
If you like LPWAN’s but are concerned about channel capacity or
possible tradeoffs between power consumption and network latency,
here is a way to accelerate LPWAN message speeds while preserving
LPWAN’s low power profiles.

28. 2. REAL-TIME DATA
1.
2.
3.
4.
5.
• WAN Endpoints send data to base station
at predefined intervals, at least 10 minutes.
• A cloud service buffers the data.
• User API is the cloud service, so user gets
data that’s at least 10 minutes old.
2.
2.
2.
2.
2.
• WAN base station can send bidirectional
queries to any or all endpoints at any time.
• Queries typically run in 1-30 seconds.
• User API can schedule queries, so user can
get data that is only seconds old.
SigFox &
LoRaWAN
Model
DASH7
Model
1.

29. 2. REAL-TIME DATA
1.
2.
3.
4.
5.
• WAN Endpoints send data to base station
at predefined intervals, at least 10 minutes.
• A cloud service buffers the data.
• User API is the cloud service, so user gets
data that’s at least 10 minutes old.
SigFox &
LoRaWAN
Model
• Mobile Asset Tracking:
10 minute old data is useless
• Public Safety Applications:
10 minute old data is useless
• There Are Multiple WAN Operators:
Difficult to know who’s cloud is
proxying the data you care about.
• If Base Station is Mobile:
Synchronized WAN model doesn’t
even work for this.
This Model Fails For…

30. HOW DASH7 QUERIES WORK
30
When an endpoint (tag) gets
a query request, the
algorithm it uses for flow &
congestion control is based
on the quality of the query.
This is a technology unique
to DASH7, which allows very
large numbers of devices to
coexist without interference.
OSI Layer
7 Application Core-apps + NDEF + UDP
6 Presentation
DASH7 Core
low power
low latency
low cost
5 Session
4 Transport
3 Network
2 Data Link
1 Physical Long range, Low Power
CoreLayersWorkTogether
forMaximumMACefficiency

31. HOW DASH7 QUERIES WORK
DASH7 Applications vs. 6loWPAN Applications
DASH7 Apps Ask:
“What are you looking for?”
6loWPAN Apps Ask:
“Who gets it?”
I need to find everyone, now, who wants to
go to floor 10.
I need data from all sensors within 5 miles
that check for vacant parking spaces.
All devices that came off the boat from
Taipei shall go to RF Channel 04 and await
further instructions.
Deliver a message to the device with
address 05:85:245:192:96:0:147:1 to turn
its lights off.
Deliver a message to the devices with
group address 124:0:8:255:37:160:0:1
instructing them to report sensor logs.
Ping device 63:102:0:80:128:0:17:44 to see
if it is still in the network.

32. HOW DASH7 QUERIES WORK
DASH7 Applications vs. 6loWPAN Applications
DASH7 Apps Ask:
“What are you looking for?”
6loWPAN Apps Ask:
“Who gets it?”
I need to find everyone, now, who wants to
go to floor 10.
I need data from all sensors within 5 miles
that check for vacant parking spaces.
All devices that came off the boat from
Taipei shall go to RF Channel 04 and await
further instructions.
Deliver a message to the device with
address 05:85:245:192:96:0:147:1 to turn
its lights off.
Deliver a message to the devices with
group address 124:0:8:255:37:160:0:1
instructing them to report sensor logs.
Ping device 63:102:0:80:128:0:17:44 to see
if it is still in the network.
If you envision a future with thousands or even millions of IoT nodes in a
metropolitan area, here is a way to query many nodes without receiving
thousands of unwanted messages from nodes that you never needed to
hear from in the first place

33. 33
REAL-TIME MAKES A
BIG DIFFERENCE
LoRaWan Haystack / DASH7
Data Access Method Periodic Beacon
Event-based
Query
Data Latency: Best Case 2 minutes 1 second
Data Latency: Worst Case 4.5 hours 10 seconds
System power for Best Case Latency
(150 mW active power)
1.05 mW 0.075 mW
Data Latency for equivalent power 34 minutes 1 second

34. 3. A “HADOOP" FOR THE IOT
• It’s a non-relational distributed database
engineered for sub-$1 microcontrollers.
• It’s built-into the data stack, so it works
directly with DASH7 networking to provide
unmatched data collection efficiency.
• Example: “Tell me the names and location of
every cow on my ranch that has not moved
in the past 8 hours”
• Example 2: “Send me a notification
whenever a 3+ year old cow moves”
34
The DASH7 file system provides a consistent data model & API allows distribution of data
and query jobs, interoperably, in real-time, across a WAN-full of Endpoints
DASH7 Data Stack
PHY/MAC/NET
Sessioning
Transport Layer
Applications
Filesystem

35. 3. A “HADOOP" FOR THE IOT
• It’s a non-relational distributed database
engineered for sub-$1 microcontrollers.
• It’s built-into the data stack, so it works
directly with DASH7 networking to provide
unmatched data collection efficiency.
• Example: “Tell me the names and location of
every cow on my ranch that has not moved
in the past 8 hours”
• Example 2: “Send me a notification
whenever a 3+ year old cow moves”
35
The DASH7 file system provides a consistent data model & API allows distribution of data
and query jobs, interoperably, in real-time, across a WAN-full of Endpoints
DASH7 Data Stack
PHY/MAC/NET
Sessioning
Transport Layer
Applications
Filesystem
A common file system for the
IoT would allow us to potentially
spider & search an open IoT.

36. 3. A “HADOOP" FOR THE IOT
• It’s a non-relational distributed database
engineered for sub-$1 microcontrollers.
• It’s built-into the data stack, so it works
directly with DASH7 networking to provide
unmatched data collection efficiency.
• Example: “Tell me the names and location of
every cow on my ranch that has not moved
in the past 8 hours”
• Example 2: “Send me a notification
whenever a 3+ year old cow moves”
36
The DASH7 file system provides a consistent data model & API allows distribution of data
and query jobs, interoperably, in real-time, across a WAN-full of Endpoints
DASH7 Data Stack
PHY/MAC/NET
Sessioning
Transport Layer
Applications
Filesystem
If you ever envisioned an IoT with endpoints that are more like smart,
data rich information servers than “dumb” terminals, here is the state-of-
the-art way of querying at the edge of the network while minimizing
network latency, channel crowding, and unnecessary power
consumption.

37. so you can either
play the lpwan
hunger games …

38. OR USE HAYSTACK & DASH7
1. Real Time Performance
2. Increased Range
3. Increased Battery Life
4. Increased Network Capacity
5. Increased Privacy and Security
6. More Use Case Options
7. Lower Costs

39. ABOUT OUR COMPANY
1. Authors of the DASH7 specification, the most advanced low power
networking protocol available. Download it here.
2. Authors of OpenTag, the open source firmware stack for DASH7 that
compiles into less than 20kb.
3. Creators of Haystack DASH7 developer tools, API’s, sample code,
reference designs, and more.
4. Creators of HayTag (in development) and other DASH7 products.
5. Founders of the industry non-profit DASH7 Alliance.
www.haystacktechnologies.com

40. SEE YOU SOON!
Contact: Patrick Burns
pat@haystacktechnologies.com
@patdash7
see
you
soon!
www.haystacktechnologies.com