LoRaWAN key to building full-stack production IoT networks
As published by NetworkWorld, July 26, 2017
Outside of the consumer market, real examples of the Internet of Things (IoT) often disappoint because in the end, they are limited by one or more of the IoT’s constraints: tens of billions of devices, cheap to acquire, cheap to deploy, security and ubiquitous connections.
I spoke with Dave Kjendal, Senet’s vice president of engineering and CTO, because he has built products and networks that meet these constraints. It was insightful because Senet has produced products using the entire IoT stack.
Senet’s evolution began in 2009 with low-cost fuel oil tank sensors communicating over the unlicensed airwaves to optimize delivery routes. The company now operates a general purpose LoRaWAN IoT network that covers one-fortieth of the United States. LoRaWAN is an implementation of low-power, wide-area networks designed to transmit small messages at a frequency of about one an hour. It serves about 55 percent of IoT WAN connectivity. It is a different technical approach than what the mobile carriers promise with 3GPP, which is yet to be standardized.
Looking back to the early 1900s at the dawn of nationwide telephone service, there were more than 2,500 telephone companies because the telephone business was an entrepreneurial business, impossible to centrally plan and execute. Entrepreneurs wired homes and business in local geographies and then were acquired and rolled up into a nationwide telephone network. The same thing happened in the early days of cable television and pagers.
LoRaWan seems to be following a similar model because it is pushing forward against the constraints limiting the IoT.
The first and foremost precipitating factor and the best indicator that a technology platform has escaped R&D and prototype stage is the availability of merchant silicon. LoRaWAN radio modules are in their third turn.
LoRaWAN roll-out and coverage
Though network components need to meet Federal Communications Commission (FCC) certification standards, LoRaWAN uses unlicensed radio spectrum, alleviating network operators from large investments in acquiring blocks of radio spectrum from the FCC. This levels the playing field for the small operator that wants to compete against a large telecom to deliver an IoT WAN. Rolling out a LoRaWAN network is accessible to a small entrepreneurial operator because of shared cell tower infrastructure. Mobile carriers do not own this infrastructure. Companies such as American Tower and Crown Castle own towers and lease space on them to operators to install their radios.
LoRaWAN operates at three power levels.
- 1 Watt or 30 dBm—for applications such as meter reading
- 0.25 Watt or 20 dBm—for applications such asset tracking trash monitoring, pest detection, tank monitoring
- 0.025 Watts or 14 dBm—for coin cell battery-powered wearable devices
Base stations or gateways can be sparsely deployed. At 20 dBm, a LoRaWAN, cell tower-mounted radio can cover a 10-mile radius and at 30 dBm up to a 25-mile radius, sometimes 100 miles. This architecture can support a sparse deployment of end devices over these distances or a dense deployment of 100,000 devices per square kilometer. There is also an indoor gateway version.
Three or four gateways cover Northeastern University, including basements and heating rooms. Senet recently demonstrated small gateways installed on the 27th floor and in the basement of the 36-floor, steel-reinforced concrete Sands hotel with two gateways that cost less than $200 in a light industrial application.
Senet operates the largest network of this kind in the U.S. The deployment model of indoor and outdoor gateways lets solution providers construct network coverage with a variety solution providers and network operators. Interoperability lets solutions providers knit together network operators into one network to meet customer needs. Senet’s coverage footprint can be combined with Comcast’s or other operators’ footprints because of LoRaWAN’s standardization.
IoT: From prototype to large-scale trial to production
Believing in IoT requires some consonant dissonance. For IoT devices to be inexpensive, they must be built at consumer scale of millions. At the same time, they need to be purpose-built, designed as semi-custum, battery-efficient devices without any unnecessary components or sensors. It is not possible to keep these two thoughts in one mind unless there is a flexible supply chain. For many customers, creating a concept prototype is possible, but they are not staffed to design the next product iteration with an integrated microcontroller, LoRaWAN radio using a developer kit.
Kjendal has seen an emerging portfolio of LoRaWAN-packaged sensors. He showed me some interesting sensors packaged for large-scale trials. One is a thumb-sized IP65 package that has a three-axis accelerometer, a magnetometer, and pressure, humidity and temperature sensors. It also has a pushbutton/LED interface and 3- to 5-year battery life that could be deployed for $20 to $25 each.
A lot of applications would use fewer sensors or require the addition of another sensor, such as a light sensor. In these cases, Kjendal would go back to the manufacturer and customize the device appropriately when scaling from 10,000 to 100,000. Other packaged sensors detect fire, smoke, water or asset tracking.
The best indicator of IoT adoption is the semiconductor manufacturers. STMicroelectronics, Murata and Microchip all build microcontrollers with embedded LoRaWAN radios to accelerate the design and manufacture of devices. Kjendal expects to see a system on a chip (SoC) sometime in the future. Similar to Qualcomm’s strategy of supplying most of a mobile phone in a single SoC Snapdragon chip, an integrated microcontroller, LoRaWAN radio with other commonly used devices integrated into a SoC would accelerate new IoT products. Kjendal says a SoC could reduce the cost to $3 per module.
Deploying a million IoT devices
In addition to the primary mission of operating its U.S. network, Senet has introduced a managed IoT network services platform for Operations Support System (OSS) and Business Support System (BSS) intended for use by network operators. The platform addresses three important business challenges.
- Cost of deployment is a significant consideration. The person deploying the end devices will be a farmer, truck driver, HVAC technician, maintenance worker, etc. who has another primary function to carry out. The task must be simple and fast, and it cannot rely on any IT skills, otherwise the deployment cost could make the project unfeasible.
- Provisioning must be secure and centrally managed. Depending on the threat assessment of the application, the platform can support a range of security protections implemented during manufacture and centrally implemented. Applications that need high levels of security can be built on hardware components with embedded trust information in the controller’s trusted element by a trusted organization. The end device built on these secure components can be cost-effectively manufactured in Asia without exposure to compromise, and it can be centrally provisioned with keys from the platform’s security controller. The platform can run independently of the network operator by the customer or solution provider. It is an extension of the eSIM model.
- Enable fast-moving entrepreneurial operators entering the market to participate in the LoRaWAN gold rush.
Using the platform, Kjendal demonstrated an over-the-air firmware update with a 100KB image. End-point devices cannot remain relevant and secure without an efficient update mechanis.
Using the entire IoT stack
Strategy and research are always interesting topics that lead to a greater technical understanding. But the road to products that customers buy and use is much more difficult to understand. My discussion with Kendjal was very interesting because he builds products using the entire IoT stack that people buy.
Senet has evolved from an oil tank monitoring device maker to a maker of gateways and end devices, to network operator and managed services platform developer. Senet also advises customers on end-device design decisions. Kendjal has experience building the full IoT stack, giving him rare valuable perspective. In New Jersey parlance, everything in IoT says easy and does hard because of the constraints of limited computational capability, long-range radios, low power, low-cost secure deployment, and managed OSS and BSS.
The most important point is IoT is not general purpose. Networks and products that succeed will be purpose-built. The choice of LoRaWAN and the unlicensed band gives IoT innovators a functional network today and gives the innovative network operators a head start on the mobile carriers that are not expected to introduce 3GPP in the licensed band until 2018 or 2019.
Senet and Kendjal also confirm the importance of the semiconductor and EDA tool makers, enabling engineers to quickly and flexibility iterate their designs from prototype to the volume production with the leanest design.