Silicon Laboratories, Inc. Rev 1.0 1
Overcoming Challenges of Connecting Intelligent Nodes to the Internet of Things
The Internet has come a long way over the last 30 years. Old-fashioned IPv4 is giving way to IPv6 so that
every device on the Internet can have its own IP address. Machine-to-machine (M2M) communication is
on the rise, enabling devices to exchange and act upon information without a person ever being involved.
The scope and scale of the Internet have changed as well: industry leaders predict that the number of
connected devices will surpass 15 billion nodes by 2015 and reach over 50 billion by 2020. The challenge
for the embedded industry is to unlock the value of this growing interconnected web of devices, often
referred to as the Internet of Things (IoT).
According to Metcalfe’s Law, the value of a network is equal to the square of the number of devices
connected to it (see Figure 1). At the edge of the IoT are the appliances and equipment we use every
day. These “things” are interconnected across an infrastructure or backbone using combinations of
ZigBee, sub-GHz, Wi-Fi or power line communications (PLC) connectivity to provide a robust bi-
directional communications link with relatively long range, low latency for fast responsiveness, low power
and a sufficient data rate to aggregate information from many connected devices. This infrastructure also
serves as the gateway to the Internet and enables remote monitoring and control of devices by other
networks, utility companies and end users.
Figure 1. “The Value of a Network is Equal to the Square of the Number of Devices Connected to It”
The majority of connected devices in the IoT, however, are nodes located at the so-called “last inch” of
the network. These nodes contain microcontrollers (MCUs), wireless devices, sensors and actuators that
provide the brains, eyes and fingers of the Internet of Things. The goal isn’t so much to enable users to
connect to all of these devices. Frankly, users don’t want to have to monitor 50+ sensors placed
throughout their homes to see if they’ve left the air conditioner on with a window open. It’s the information
these devices gather that’s important, as well as the ability of machines to communicate among
themselves and make decisions so we don’t have to (see Figure 2).
Silicon Laboratories, Inc. Rev 1.0 2
Figure 2. Home Area Networks Often Contain Numerous Connected Devices
The challenges of implementing connected device applications for the IoT are quite different from those
associated with traditional network edge devices. Lighting and appliance OEMs, for example, will need to
bring in new networking, wireless, and embedded software technology beyond their core competencies.
They can either develop these technologies themselves or partner with companies that have already
created products that can be easily introduced into systems.
Because only a small amount of data typically needs to be transferred between these devices, a robust,
cost-effective wireless protocol like ZigBee is ideal. Power efficiency is critical since these devices are
often not connected to power and have to operate using energy harvesting sources or a single battery for
several years without maintenance or battery replacement. Developers also need to consider factors,
such as cost, component count, microcontroller performance, system size, standards, interoperability,
security, ease-of-use and in-field troubleshooting. Finally, software is required to bridge devices,
aggregate sensor data and present information to end users in an intuitive way via displays or over the
Internet to their computers, tablets or smartphones.
Envisioning the IoT
Smart meters represent a prime example of a high-profile Internet of Things application. Rather than
simply measuring power consumption, smart meters enable utility companies to communicate in near real
time with consumers or through opt-in programs and proactively shut down the operation of heavy load
appliances, such as air conditioners, during peak-demand times. The result is a lower electricity bill for
consumers and a shift of loading so that utility companies don’t have to invest in new power generation
sources for the few days in a year when supply is challenged by demand.
Smart meters are just one aspect of the emerging smart home. In addition to sharing computing files and
multimedia content, connected home networks enable a wide range of security, monitoring and
automation applications comprising intelligent lighting, smart appliances and other devices. The
availability of even a few sensors – temperature, motion, humidity, light, glass breakage – enables a
powerful mesh network that extends the capabilities of all devices connected to it. In fact, the IoT can
provide significant benefit to industrial automation, lighting control, home/building automation, security
and monitoring, health and fitness, and agricultural applications, to name a few (see Figure 3). The IoT
also provides new ways to interact with devices. The term “app-cessory,” for example, has already been
Silicon Laboratories, Inc. Rev 1.0 3
coined to refer to applications on a user’s smartphone that can communicate and control sensors and
lights in the home and business.
Figure 3. In the Next Few Years, the IoT will Connect Tens of Millions of Devices across Numerous
Industries Using the ZigBee Protocol
One application of Metcalfe’s Law involves devices that are not practical to install on their own but can
add tremendous value when they are able to leverage existing infrastructure. Consider the reduction of
“vampire power” within the home and business. Vampire power refers to devices, such as TVs and set-
top boxes, that consume power when they are not being used. Experts estimate vampire power to
represent between 7-15 percent of total electricity used in the home. Installing motion sensors to detect if
a person is in a room so that power to the TV or set-top box can be turned off would clearly be cost-
prohibitive. However, when a TV can leverage motion sensors already installed as part of a home security
system, then vampire power can be managed in a cost-effective manner.
The sensors of a home security system can be used for a wide range of other applications as well. For
example, the lighting system can be tied in to turn lights on when a person enters a room and
automatically turn them off when no one is present. Existing systems can also be extended. For example,
a low-cost humidity sensor can be added to a security system to automatically turn on and shut off an
exhaust fan after a shower. Such interactions between different systems bring a number of benefits to
Higher Efficiency: When connected to the IoT, devices can determine the best time to operate; i.e., a
clothes dryer can wait until after peak demand hours to operate using lower-cost electricity.
Proactive Usage: Today, users can set the air conditioner to run for when they plan to get home from
work. If they are late, the system will operate with no one home. Smart home systems, such as Iris from
Lowes and smart energy systems from AlertMe, enable remote control of climate control systems so end
users can alert their homes to make shifts in usage for higher operating efficiency. These systems can
also communicate with homeowners when required through text messaging.
Proactive Maintenance: Intelligent devices can monitor their own operating health and notify users or
OEMs of potential issues before they result in downtime. For example, a dishwasher may exhibit a
recognized wear pattern that leads to failure within, say, three months, enabling an OEM to automatically
Silicon Laboratories, Inc. Rev 1.0 4
update the system with new features and algorithms that increase efficiency and drop operating costs.
This can also reduce the number of warranty service calls for OEMs.
Single Control Interface: Since users can use their own devices, such as smartphones, to manage the
network, it becomes possible for a single application to control devices rather than requiring users to learn
a different user interface (UI) for every new appliance or node added to the network. Note also that, for
many applications, implementing a display is not cost-effective. On a washer, for example, a robust and
more expensive display would be required to provide sufficient durability to handle the shaking of the
machine. Another issue is that displays are typically out-of-date by the time the appliance reaches the
store. For these and other reasons including reduced system cost and complexity, OEMs are exploring
ways to enable end users to manage their own displays. The Nest Learning Thermostat, for example,
enables homeowners to remotely program climate controls from any Internet-connected device, such as a
Ease-of-Use: When devices can be managed over a network, users have the ability to control the network
from anywhere they want, using the applications they want. Troubleshooting is greatly simplified as well.
For example, instead of a dishwasher lighting up several LEDs to signal an error code, the device can
clearly describe any operational failures or issues.
Note that these interactions are from machine-to-machine and do not require user involvement. Rather
than each system working independently and making decisions with limited data, the IoT enables
systems to share information to greatly expand their capabilities and value beyond their original design.
The convergence of various applications enables all of them to work better.
The power of Metcalfe’s Law means opportunities for companies in every industry. While a security
company could expand its reach into the lighting and home automation markets, it could instead partner
with established lighting and home automation vendors to create value-added services. This is the power
of an ecosystem. The Internet of Things enables electronic component suppliers, software vendors,
OEMs and service providers to focus on their core competencies and leverage the strengths of
partnerships to create compelling applications for consumers.
Interoperability through Standard Protocols
For the IoT to work, all devices must be able to connect seamlessly. However, there is no one wireless or
wireline technology that can efficiently serve across an entire network. To develop cost-effective products,
engineers need to be able to select the optimal communications channel and protocol for their
application. As a result, the IoT will be based on a variety of standard and proprietary protocols.
For devices to be able to reach out across the Internet, they will need to support IP somewhere along the
communications channel. At the edge, however, IP can be a rather full-featured protocol with a great deal
of overhead and cost for simple applications. Similarly, while Wi-Fi is ubiquitous, it consumes too much
power for devices restrained to battery or energy harvesting power sources.
Connected devices need to be able to use protocols, such as ZigBee and 6LoWPAN, that are lightweight
and have a data rate that reflects their requirements. Devices that connect to the IoT through a
centralized controller can even employ proprietary standards given that their data is aggregated and
converted to a standard format before being passed out onto the Internet via a gateway device.
The ideal combination of radio technologies and protocols depends upon the specific application. Today,
Wi-Fi is the appropriate technology when high data rates are required, such as when transporting video.
For low-bandwidth applications that do not require direct user interaction, 2.4 GHz ZigBee and sub-GHz
technologies present a lower power wireless link that is much more easily integrated into embedded
systems. For simple applications, such as garage door openers or systems requiring long-distance
connectivity like irrigation systems, using a sub-GHz radio is likely the optimal approach. If two-way
communication, security or a large number of devices need to be connected in a mesh network, ZigBee
offers a robust implementation.
Silicon Laboratories, Inc. Rev 1.0 5
Employing a mesh topology is ideal for many Internet of Things applications. Consider a home lighting
system where the number of nodes can quickly exceed 30 lights and sensors. Whereas a Wi-Fi router
may not be able to provide whole-house coverage, a mesh topology enables robust coverage for every
location within the house with the lowest per-node cost. In addition, meshes can automatically configure
new devices so that they leverage usage patterns that the system has already learned. Scalability is an
important factor as well. Bluetooth, for example, is limited to just seven devices on a network and Wi-Fi to
32. Networks based on Silicon Labs’ EmberZNet Pro provide self-configuring and self-healing mesh
connectivity that can be extended to interconnect hundreds or potentially thousands of devices on a
Achieving Ultra-Low Power Efficiency
Because last-inch devices typically perform limited tasks, they tend to have fairly simple architectures
focused on basic data collection, calculation and connectivity functionality. Whether such a device needs
an 8-bit or 32-bit microcontroller depends primarily on the types of calculations the device needs to
perform. The wider bus and advanced peripherals of 32-bit MCUs also enable substantially faster data
movement and computational power than 8-bit MCUs, so devices can return to sleep faster for better
For many last-inch applications, such as motion and light sensors placed throughout a house, the cost of
installing new wiring to power these devices is prohibitive compared to the cost of the device and the
function it is to perform. As a consequence, these devices must offer superior power efficiency so they
can operate using a battery or harvest energy from their environment. In addition, these devices must be
easy to install, even in difficult-to-reach spaces, and they must be able to operate for years without
requiring battery replacement or other servicing.
To meet the power requirements of last-inch devices, MCUs must support ultra-low-power operation while
enabling ubiquitous control and connectivity. To achieve this exceptional power efficiency, MCUs must
integrate a variety of advanced capabilities. An on-chip, high-efficiency dc-dc buck converter, for
example, can enable higher efficiency while allowing devices to operate all the way down to the lowest
usable voltage of the batteries. A 32-bit microcontroller with an integrated dc-dc buck converter, such as
Silicon Labs’ 32-bit SiM3L1xx Precision32™ device, can achieve 40 percent lower active mode power
compared to a similar microcontroller without a buck converter, as shown in Figure 4.
Figure 4. An Integrated DC-DC Buck Converter Can Reduce Active Power by Up to 40 Percent
Silicon Laboratories, Inc. Rev 1.0 6
MCUs used in devices at the edge of the IoT ideally support multiple power domains, enabling
peripherals to operate autonomously at different frequencies with the CPU powered down. For example,
direct memory access (DMA) can be used to collect sensor data and wake the CPU only when there is a
full buffer of data to process. This results in a greater sleep-to-wake ratio and higher power efficiency.
These architectures are also highly specialized to transition in and out of standby faster to reduce the
power wasted while the CPU wakes.
Silicon Labs’ SiM3L1xx 32-bit microcontroller family, for example, features a dedicated, programmable
Data Transfer Manager (DTM) hardware block that enables the embedded designer to chain together a
complex set of tasks that execute autonomously without relying on the MCU core. In these instances, the
core is kept in its lowest power state until all tasks have completed. The DTM is especially useful in
wireless data transfers. In wireless systems, for example, raw data is processed through multiple
operations before being delivered to the radio for transmission. The microcontroller must encrypt the raw
data, add error correction, encode the packet and pass the packet to a serial interface in one or more
bursts. This process is shown in Figure 5.
Figure 5. Silicon Labs’ Data Transfer Manager Enables Autonomous Data Transfers of Radio
Packets without CPU Intervention
Application-specific circuitry can further offload the CPU. For example, an on-chip pulse counter for utility
meters can efficiently measure pulse trains or fluid flow as a background operation while integrated
capacitive touch-sensing capabilities can reduce active time power consumption for devices with a UI.
The availability of hardware-based accelerators that speed processing and decrease active current, such
as dedicated packet processing engines (DPPEs), can vastly improve RF message processing while
allowing the CPU to remain idle during transactions.
Accelerated Software and Application Design
Software plays a critical role in enabling the features and capabilities required to build out the IoT.
Whereas hardware provides the foundation for connectivity, software enables the underlying M2M
interactions that ensure that devices operate in a reliable and robust manner regardless of the operating
Consider how the quality of a wireless connection is highly dependent upon the operating environment;
i.e., every cell phone user has experienced the need to move around inside a house to get a better signal.
Devices such as thermostats are not going to move even an inch to get an improved signal. Latency also
comes into play; after 100 ms without a response, most users will tend to press a button again.
Software is what makes wireless networks robust. It ensures that messages have been received and
acted upon, such as a light actually turning on. Software also enables developers to implement greater
intelligence and flexibility into devices so they can identify problems, raise exceptions and potentially
resolve issues without the need for human intervention.
Developers are also able to implement advanced functionality through software. For example, while it is
useful to be able to turn on a light remotely, it is even more useful when the light can tell a user that the
Silicon Laboratories, Inc. Rev 1.0 7
bulb needs replacing. Software extends the possible range of autonomous control to further improve
efficiency and convenience. Consider that, with an intelligent wireless sensor network, a home could
determine when no one is home and power down all electronic devices. The result of such a simple
change of operation, when multiplied over hundreds of millions of households, is considerable power
Software also serves as the bridge between networks and controllers. For example, some users will
prefer to control appliances and lights from their smartphones, others from their tablets. Software enables
each person to use the device and interface of their choice. Flexible, easy-to-use software also provides a
means for OEMs to differentiate their products from their competitors.
The reality is that we are shifting to an app-driven world. MCUs, sensors and wireless links may form the
foundation of the IoT, but the real innovation will occur in software technology. After all, what is useful is
not the data from sensors all around our homes but rather what we can do with it.
The key to this is interoperability and open standards, which enable a wide range of devices to
communicate with each other. This capability of device-to-device communications enhances the value of
the network, and, once the network infrastructure has been created, more information and intelligence
can be obtained at negligible incremental cost. To achieve this level of sophistication, software needs to
abstract specific hardware details by providing a common application layer that can be shared between
devices and applications. In this way, the underlying technology used to transport data becomes
irrelevant, freeing the developer to innovate around the IoT application.
For example, the global ZigBee standard, pioneered by the ZigBee Alliance, gives connected device
manufacturers a straightforward way to develop standards-based products capable of interoperable M2M
communications. ZigBee standard profiles, such as ZigBee Smart Energy, ZigBee Home Automation,
ZigBee Building Automation and ZigBee Light Link, provide interoperable platforms that simplify the
development of IoT applications for smart homes and commercial buildings, intelligent lighting control
systems, smart meters and in-home energy monitoring systems.
Faster Time-to-Market for IoT Applications
To help engineers bring their own IoT devices to market faster, semiconductor suppliers must offer a
diverse range of advanced design tools, such as application libraries for accelerating the implementation
of key functions, production-ready sample applications, firmware development tools, complete
communication and radio stacks with built-in security, and simple demonstration applications that show,
for example, how to connect a smartphone to a last-inch device over the Internet.
Today, development tools are available that provide a macroscopic view of the entire network from a
single console and create a back-channel link to facilitate troubleshooting and tracing of packets across
the network. Specialized wireless development tools are also available to enable developers with little to
no RF design experience easily create efficient, robust and cost-effective ZigBee and sub-GHz wireless
applications. With the availability of a wide variety of development boards for evaluating the connectivity
and performance of various protocols, engineers can simultaneously design and debug application code
and firmware, begin RF design and optimization, and finalize network and protocol stack development
while hardware prototypes are still under development.
Silicon Labs’ EmberZNet PRO protocol stack, for example, provides a ZigBee-compliant software solution
for IoT applications. Deployed in more wireless networking products than any other ZigBee stack,
EmberZNet PRO software provides enhancements for robustness and ease-of-use that maintain
complete ZigBee compliance while offering developers an edge over standard ZigBee PRO feature set
implementations, especially in larger networks and more challenging environments.
The EmberZNet PRO stack is complemented by an Ember ZigBee development environment that
provides sophisticated visualization and debugging tools to shorten design time. For example, the Packet
Trace Port and Ember Debug Adapter provide visibility into every bit transmitted/received by each node in
Silicon Laboratories, Inc. Rev 1.0 8
the network through a standard Ethernet connection. The Ember Desktop Network Analyzer enables
rapid visualization of network activity and network debugging. The easy-to-use Ember AppBuilder tool
provides the fastest path to ZigBee-certifiable wireless networking products using ZigBee standard
application profiles. Full application templates for the ZigBee Smart Energy, Home Automation and Light
Link profiles are available from Silicon Labs.
The value of connecting devices to the Internet and having them seamlessly communicate with each
other independent of human intervention is no longer under debate. The IoT will continue to open new
markets and drive new applications and opportunities for OEMs and application developers across all
industries. Nor is there any question about whether the IoT is going to happen given the rapid expansion
of applications, such as smart meters and smart home appliances. The IoT has become a tangible reality
with commercially successful deployments in several markets, including connected home and green
What many OEMs and their suppliers want to know is when the Internet of Things is going to grow out of
its infancy and achieve the critical mass necessary to become a 10 M+ unit market. With the availability of
the fundamental technologies, products, software and tools necessary to create efficient, ultra-low power
devices for the last inch, it is clear the answer is now.
# # #
Silicon Labs invests in research and development to help our customers differentiate in the market with innovative low-power, small-
size, analog-intensive, mixed-signal solutions. Silicon Labs' extensive patent portfolio is a testament to our unique approach and
world-class engineering team. Patent: www.silabs.com/patent-notice
© 2012, Silicon Laboratories Inc. ClockBuilder, DSPLL, Ember, EZMac, EZRadio, EZRadioPRO, EZLink, ISOmodem, Precision32,
ProSLIC, QuickSense, Silicon Laboratories and the Silicon Labs logo are trademarks or registered trademarks of Silicon
Laboratories Inc. ARM and Cortex-M3 are trademarks or registered trademarks of ARM Holdings. ZigBee is a registered trademark
of ZigBee Alliance, Inc. All other product or service names are the property of their respective owners.