A look at the shifting architecture of embedded systems (particularly in light of IoT and wearables), how ecosystems are built and how the industry is evolving

1. Embedded Platforms and Ecosystems
Dan Ledger
@dledge
dan@endeavourpartners.net

2. Embedded Platforms and Ecosystems
13th Aug 20152
Platforms are a powerful tool for growth and innovation
in technology & information companies
Platforms enable economies of
scale and scope by leveraging
shared assets
Platforms enable diversity and
evolution while reducing costs
Platforms enable ecosystems
that drive innovation
1
2
3

3. Embedded Platforms and Ecosystems
13th Aug 20153
The right platform strategy allows semiconductor companies to
increase the value and appeal of their products, increase cross-
sales, and drive lock in
A
B
C
A platform provides a
set of core and requisite
set of functionality upon
which others can build.
A platform may be
provided by one
company or through
close partnership of a
few companies.
An ecosystem of partners
extends the value of the
platform, and may help
open up new types of
customers. These
partners can provide a
variety of capabilities as
well as services.
existing customers new customers
Example
A
Product strategy
existing customers
Platform strategy
In a traditional product
strategy, a supplier may
build a stand-alone
offering, or work with a
few captive partners on
a complete solution
Intel is launching an IoT platform offering with
Wind River and McAfee (both of whom Intel
acquired) which will leverage Wind River’s
ecosystem
platform ecosystem

4. Embedded Platforms and Ecosystems
13th Aug 20154
Platform strategy is becoming particularly important in connected
applications where architectural innovation is taking place
In the past
(monolithic, integral)
Emerging
(distributed, modular)
Stand-alone system
containing all capabilities
Functionality implemented
at the node / edge
Functionality
implemented in the
cloud
Exposed capabilities
(via APIs) from others
Emergence of tens of thousands of
APIs now allow companies to
inexpensively offload processing
to others that is outside their areas
of specialization
Several device / data
management platforms allow
companies to more easily
manage their nodes
{
Ubiquitous LTE and high
speed networks enable
high-speed, low-latency links
to the cloud / infrastructure
All capabilities
implemented
within a rigid
stand-alone
system {
Capabilities
distributed to
improve system
economics,
performance and
scalability

5. Embedded Platforms and Ecosystems
13th Aug 2015
Sensors /
actuators
5
This shift is enabling several new system architectures
in which functionality is broadly distributed
Gateway
Gateway
Sensors /
actuators
Sensors /
actuators
Sensors /
actuators
Sensors /
actuators
Applications Applications Applications Applications Applications Applications Applications
Device and data management platform(s)
Client
Client
Client
e.g., IP Camera e.g., Local sensor network e.g., Remote location sensor
Third party capabilities

6. Embedded Platforms and Ecosystems
13th Aug 20156
This shift in architecture is introducing several new
challenges for OEMs as well as system integrators
• Security
• System complexity
• Provisioning
• Device management
• Firmware updates
• System integration
and interoperability
• Lock-in
• Latency
• Reliability

7. Embedded Platforms and Ecosystems
13th Aug 20157
There are several emerging standards that are helping
to address these issues, but standards are not enough
Device
management
Session
Application
Datalink
Network
Transport
Physical
WiFi Wired802.15.4
Bluetooth
Smart
Low power
cellular
(e.g. Sigfox)
3G / 4G
cellular
DASH7
OMA DM
ZigBee IP /
Pro
Z-Wave
INSTEON
EnOcean
6LoWPAN
IPV6 / RPL
UDP TCP SMS
Thread
(Google, NEST,
ARM, Samsung,
etc.)
ZigBee
ZigBee
AOs
AllJoyn
(Qualcomm, LG, Sharp,
Sony, Cisco, HTC, etc.)
Physical
Web (BT)
(Google)
IPV4/IPV6
CoAP
OMA LWM2M
TLS MTLS
Proprietary
MQTT HTTP
ISPO App Objs
More widely deployed standards

8. Embedded Platforms and Ecosystems
13th Aug 20158
OEMs are increasingly relying on operating system that
contain building blocks required for distributed architectures
Inhouse/custom
Android
FreeRTOS
Ubuntu
Debian (Linux)
MS Windows Embedded 7
Micrium uC/OS II/III
MS Windows Compact
Wind River (VxWorks)
TI DSP/BIOS
Angstrom (Linux)
Keil (RTX)
Freescale MQX
QNX
Red Hat (Linux)
TI RTOS
0% 7.5% 15% 22.5% 30%
8%
5%
5%
7%
6%
5%
7%
7%
7%
12%
9%
10%
14%
26%
27%
15%
4%
5%
3%
5%
4%
4%
7%
7%
8%
6%
11%
8%
13%
13%
16%
24%
Was using in 2013
Considering in 2015
Source: UBM 2014 Embedded Market Study

9. Embedded Platforms and Ecosystems
13th Aug 2015
As the system architectures evolve, companies are partnering
further up the stack to offer more complete platforms
9
Component
Supersystem
(node + infrastructure)
Algorithmic
building blocks*
H
C
System
(node)
Architectural
(OS and frameworks)
Architectural
(device & data
management platforms)
APIs & exposed
capabilities*
Hardware
components
* There are hundreds of companies in these segments, this particular subset contains examples related to health

10. Embedded Platforms and Ecosystems
13th Aug 201510
Semiconductor suppliers are beginning to partner higher up
the stack to provide a more compelling and complete offering

11. Embedded Platforms and Ecosystems
13th Aug 201511
Semiconductor companies are experimenting with new ways
to drive development on and adoption of their platforms
System-level
platforms
Form factor
reference design
Complete end
product platform
Freescale Wearables Reference
Platform (WaRP)

12. Embedded Platforms and Ecosystems
13th Aug 201512
Companies must not only understand the needs, but also
create a set of beliefs across their potential customers and
ecosystem partners
Partners / developers Customers / OEMs
Needs Requisite Beliefs
• A core platform that
contains requisite
functionality
• Good tools,
documentation and
support
• A path to reach customers
• Co-marketing
opportunities
• The platform will be
important in this market
• This platform provider is
committed to helping
partners succeed
• Developing for this
platform is the optimal use
of time / $
• I will be able to monetize
this investment
• I’ll get the support I need
Needs Requisite Beliefs
• A robust, complete and
competitive set of
functionality
• Good value (cost / benefit)
• Low execution risk
• An ability to differentiate
an end product
• The platform will be
important in this market
• The company providing
this platform is committed,
and I’ll be able to rely on
this platform for future
products
• We’ll be able to build what
we need to
• We’ll get the support we
need

13. Embedded Platforms and Ecosystems
13th Aug 201513
Intel, for example, has struggled for years to develop ecosystems around
several of its products despite strong technology and significant investment
Source: UBM 2014 Embedded Market Study
Texas Instruments
Freescale
Microchip
ARM
Atmel
STM
Intel
Xilinx
NXP
Renesas
Other
0% 10% 20% 30% 40%
2012
2013
2014
Which suppliers has the best ecosystem (write-in)?

14. Embedded Platforms and Ecosystems
13th Aug 201514
Amongst other things, a strong theory of mind is particularly essential
in order to plan, develop and sustain successful ecosystems
Theory of mind is the
ability to attribute the
mental states (belief,
desire, intent, etc.) of
other individuals.
First order theory
of mind
“I believe that Barry
thinks this product is
too expensive.”
“I believe that Barry
wants Judy to think
that this couch is
perfect.”
Second order
theory of mind
Nth order theory
of mind
A strong theory of mind also allows one to
anticipate mental states based on a hypothetical
set of future events
A strong theory of mind provides second, third and
fourth order attribution of mental states
“I believe that Barry would think
Judy would be very displeased if I
were to miss the dinner.”
“I believe that Barry thinks this
product is too expensive.”
Apple succeeded in building an ecosystem because it had a strong theory of
mind; Microsoft and Nokia failed because they did not.