Need to implement wireless device networking? Here’s what to look for

Over the last two decades, billions of people have connected themselves to the Internet using computers, and more recently, mobile phones.  This communication revolution is now extending to objects as well. This revolution goes by many names, from “machine-to-machine” (M2M) communication to the “Internet of Things.” Whatever the name, the vision of a device network is the same: to connect objects so that they have the information to function optimally (1). This vision includes:

* Smart electric grids that automatically adapt themselves to fluctuating load conditions and drive lower energy use by spreading demand.

* Gas and water distribution systems with automated metering and leak detection.

* Buildings that can adjust to changing temperatures in order to save energy.

* Shipping containers that track temperature, humidity, location, and if they have been tampered with as they travel across the world.

Figure 1 lists examples of more applications targeted across various industries. Though the data throughput requirements for these devices are often relatively low, there are significant challenges in connecting them:

* They are widely distributed across large geographic areas.

* They are often located in difficult radio environments, e.g. in basements, below ground, and in shielded environments.

* They require latency in near real-time, i.e. on the order of seconds, less than one minute.

* They require years of dependable battery operated power.

* They require an extremely secure network due to their critical nature.

Figure 1.  Target Applications

Due to these challenges, the majority of these devices remain unconnected. Wired solutions are prohibitively expensive, except in the special case of electric utilities, which have their assets connected by power lines.  But even there, power-line communication (PLC) has some serious drawbacks with huge voltages, signal attenuation through transformers, and an extremely challenging noise environment.  By and large device networking solutions must be wireless.

Only recently has wireless technology been developed to address the particular needs of this segment.  In this article, we will highlight the important factors to consider when choosing a wireless communication technology for device networking.  Of course, the communication link is only one piece of a complete device network, but it is the foundation—the base of the pyramid as illustrated in  Figure 2.

As the foundation, it is an extremely important piece to get right.  In any communication network, the physical (PHY) and medium access control (MAC) layers are closest to the laws of physics. Consequently, it is virtually impossible to correct inadequacies in the PHY/MAC at higher layers without sacrificing other fundamental aspects.  For the utility space in particular, the stakes are very high. The selection of a non-suitable technology will result in increased expense and degraded reliability for the envisioned applications. The incorrect choice could threaten the existence of a substantially larger ecosystem that uses the foundational technology.  Fortunately, the performance of these lower pieces of the pyramid can be analyzed objectively, and unlike higher layers, clear decisions based on objective criteria can be made.

Figure 2. The PHY and MAC layers are the foundation of any wireless communication technology

Wireless Performance Metrics

This article focuses on five key performance metrics to consider when evaluating a wireless solution for device networking:

* Coverage: The range of a wireless link to reliably connect devices. The range is measured in meters from the access point (AP) and varies depends on the RF propagation environment. Factors, such as terrain, clutter type (urban, suburban, rural, etc.), interference, and local output power regulations determine the range capabilities of a specific radio.

* Capacity: The amount of data from device endpoints an AP can simultaneously serve. The capacity is measured as the aggregate application throughput (kbps) at the AP.

Coverage and capacity must work in parallel to achieve a highly functioning network. They work in conjunction to drive the ratio of APs (or network infrastructure points1) to device endpoints. Providing coverage for many devices without proper capacity is pointless; conversely, excess capacity without coverage to reach devices is equally pointless. It is the optimal mix of the two that delivers cost effective performance for a wireless network.

* Coexistence: The ability of a wireless technology to coexist with other devices that can cause significant and dynamic interference. This is particularly important when operating in unlicensed bands.

* Power Consumption: The ability to support a long battery life. Many devices, such as gas and water meters, are not continuously powered. Connecting these devices with extremely long battery life wireless modules, on the order of years, is important in creating a cost-effective connectivity solution.

* Security: The ability of a wireless system to resist malicious attacks from cyber criminals. To prevent malicious attacks, the device networks, particularly those focused on critical infrastructure, require proven and robust cyber-security across the network.

Coverage

The primary driver of coverage is link budget. Link budget measures the ability of a wireless system to overcome obstacles to close a link. Link budget is driven by many factors, but at the highest level, there are three primary factors that contribute to link budget: transmit power, antenna gain, and receiver sensitivity.

In all radio frequency bands, especially unlicensed ones, regulation limits the Effective isotropic Radiated Power (EiRP), which ultimately limits transmit power and antenna gain. This leaves receiver sensitivity as the differentiator. Receiver sensitivity reflects the radio's ability to detect signals—the lower the sensitivity, the better.  This is particularly important in the device networking space to ensure that the signal can penetrate into difficult radio environments.  A good rule of thumb is to compare the receiver sensitivity to the thermal noise floor. A strong radio will have the ability to operate at negative SNRs, well below this floor.

Wireless meshing can be used to extend the limited range of weak radios. When deployments are dense and in favorable RF environments, such as a suburban neighborhood, wireless mesh can be successful2. When deployments are sparse or not in favorable RF environments the mesh architecture fractures into many clusters, requiring extensive APs to ensure reliable connectivity. Despite its success in certain environments, mesh has significant limitations and is fundamentally a “patch” to the limited link budget of a radio. The better solution is to deploy a wireless technology that is purpose-built with a massive link budget that can reliably deliver extensive coverage.

Capacity

In discussing capacity, it is key to differentiate between the single-link throughput, from a single endpoint to an AP, and the aggregate capacity of the AP serving many endpoints.  The single-link throughput of a technology could be modest, but if there is a good multiple-access scheme, the aggregate capacity could be extremely high.  Since most devices have low throughput requirements (e.g. grid transformers do not need to stream video) the aggregate capacity is the metric that matters.

Many wireless technologies quote a raw PHY-layer data rate as if it was their aggregate capacity.  But that number often has no bearing on the aggregate capacity of interest – the application-layer data rate able to be sustained at the AP.  This number is often much lower because of MAC overhead and poor multiple access techniques. In addition, wireless systems cannot be operated at capacity in steady-state. There must be some margin to account for unpredictable traffic fluctuations. The network topology and multiple access technique drive this required margin.  With a star topology and a good multiple access technique, an AP can be safely operated at ~60% capacity.  With a mesh topology and ALOHA random access, an AP can only be operated at ~20% capacity due to the inherent instability of the access method.

In short, when comparing different technologies, it is key to focus on the application-layer capacity at the AP – taking all of the effects above into account. 

Coexistence

Like all wireless devices, device-networking radios must be robust to interference and a propagation environment that is in constant flux. Robustness has a number of key benefits.  It enables deployment in the unlicensed, but noisy, ISM bands (like 900 MHz and 2.4 GHz) where the transmitters are lightly regulated. Licensed spectrum is expensive and controlled by the vendor that owns it.  In many situations, it is much more cost effective and flexible to work in unlicensed spectrum.

In addition, robustness enables a radio take advantage of elevated AP locations, such as those located on mountain top antenna farms, mobile operator RF towers, and tall buildings.  This provides far better coverage from a single AP.  Less robust technologies cannot use these types of sites due to the interference they receive from co-located transmitters and the surrounding areas.

A strong radio will have interference mitigation techniques built in at every protocol layer. At the PHY-layer, techniques such as direct sequence spread spectrum (DSSS) that can operate at negative SNR are particularly strong because they are designed to withstand far more co-channel interference. At the MAC, it is imperative that an acknowledged point-to-point data transfer between the endpoint and the AP is implemented.  This usually includes an Automatic Repeat Request (ARQ) mechanism that ensures reliable delivery even under large packet error rates.  The protocol makes the underlying wireless medium, with all its interference and channel variation, seem like a wire to the higher protocol layers. 

Power Consumption

The device networking space covers both continuously powered devices, such as electric meters, and battery-powered devices, such as fault circuit indicators (FCIs).  From a business perspective, it is extremely desirable to have a single network that supports both, as it would be prohibitively expensive to install and manage two separate networks. 

A strong device networking technology should be configurable to provide a low latency, highly responsive connection for continuously powered devices and an extremely power-efficient connection for battery-powered devices.  To be cost effective, applications such as water meters, gas meters, and FCIs, should be able to achieve a battery life of greater than 10 years.

Power efficiency is driven by the wireless protocol.  The key to efficiency is to minimize the time and power spent transmitting and receiving radio waves.  This can be done using various techniques including fast acquisition – to reduce the amount of time it takes to find the network and adaptive modulation – to minimize the amount of time needed to transmit the packet.  In addition, the protocol should be designed such that the endpoint is in a low power “deep sleep” mode most of the time.

Security

Many of the target applications for device networking are critical infrastructure endpoints. Once the endpoint devices are connected, they can become the targets of hacking and cybercrime. They require a secure network that is built using proven algorithms.  A strong network should use a comprehensive defense in depth strategy to deliver this information security, including:

* Prevention mechanisms: Provide access control, mutual authentication, confidentiality, and high availability.

* Detection mechanisms: Identify system breach attempts and alert operators.

* Recovery mechanisms: Ensure the system degrades gracefully and continued operation even while under attack.

A secure network should provide the key security guarantees as in Figure 3 below, preferably using NIST approved security algorithms that have greater than 20 years of life. 

Figure 3. Key Security Guarantees

Conclusion

There is a confusing array of vendor choices when evaluating wireless device networking solutions including:

* Cellular

* Unlicensed wireless mesh radios

* Lightly licensed narrow band radios

* Unlicensed wide band radios.

Many of these technologies were originally designed for applications that are significantly distinct from the requirements of device networking.  Consequently, they have serious limitations in scaling to a low bandwidth, widely distributed network with potentially tens of thousands of hard-to-reach devices.

For example, cellular technology is evolving in the opposite direction of what is required for device networking. Consumers are clamoring, and paying, for costly endpoints, such as smart phones and tablet PCs. This is driving the cellular industry toward high bandwidth, high cost, and power hungry wireless communication technologies. This clearly goes against the grain of many performance aspects that are essential for device connectivity: Low-cost infrastructure, efficiency for small payload data transactions (as opposed to streaming video), and coverage of locations where consumers do not frequent – e.g., underground, rural.

Due to the unique challenges device networking presents, a successful wireless technology for this space must be purpose-built.  Adapting a technology designed for a different purpose is like putting a square peg in a round hole -- leading to fundamental weaknesses. Fortunately, there are a number of companies that have been developing technologies purpose-built for this space.  When choosing a technology, however, you should judge it objectively using the five wireless performance metrics outlined above.

¹ Network infrastructure is required to support the network and route the data back and forth. The term, Access Point (AP), is used for the ULP system. Base station or gateway may be used to refer to other systems.

²This has yet to be proven in regions with lower EiRP limits than in the United States.

About the Author

Sourav Dey is a staff systems engineer at On-Ramp Wireless, where he develops the M2M technology for the wireless communication backbone of the smart grid.