How Did 3 Engineers Tape Out an ARM IoT Product in 3 Months?

The Internet of Things (IoT) is the hottest buzzword (buzzphrase?) in electronics right now. Everyone's definition is a little different, and each time I see a presentation on IoT the number of units goes up by a few more billion. Probably the most widely used number is Cisco's 50B devices in 2020. Often, devices defined as now in the IoT market were previously counted in other markets such as automotive or medical, so the definition and the number of units vary. I think that three things characterize an IoT device: one or more sensors, some sort of microprocessor to run the software, and connectivity to allow the device to access the internet. Many IoT devices will be designed to operate for a long time without requiring the battery to be changed, perhaps even for the lifetime of the product. This means that low power is going to be critical. A typical IoT device will not live in isolation. Usually there will also be some sort of cloud-based back end to store the data uploaded and do any heavy computational lifting. It is easy to dismiss IoT as a lot of hype, but it is clear that there really will be lots of intelligent devices of one sort or another in our homes and on our persons.

One market that will be important is industrial IoT. The value proposition in this market is clearer, typically a cost saving. There is not the same fashion aspect as there is in the consumer IoT market, an aspect that makes predicting success or failure much harder. William Goldman is famous for a quote about Hollywood, "nobody knows anything...not one person in the entire motion picture field knows for a certainty what's going to work." I think this could equally well be applied to consumer IoT applications. The markets are very fragmented and it isn't really possible to find out what we, the public, will buy other than building something and trying to sell it. You can see this, for example, in the smart watch segment where all sorts of different approaches are being taken. Some IoT devices will be built by integrating standard products, such as the Fitbit fitness tracker, and others will be built around an SoC, such as the Apple watch. Whichever approach is being used, it is important to be able to get to market quickly. Since the first product to market is unlikely to be the perfect match for the marketplace, it is also important to be able to iterate the design fast, too.

An IoT device is almost certainly not a huge SoC on a cutting-edge process node. A more likely profile is a clock speed of <200MHz, less than 500K gates, on 55/40 ULP or a similar process, fewer than 100 I/Os. Total die size, 5mm².

ARM has the leading position in embedded microprocessors and so it naturally expects that it will have a large share of IoT devices as the market develops. To reduce the time to market of ARM-based IoT devices, ARM and Cadence have collaborated to produce an IoT reference system. This is the first result of the Cadence/ARM strategic IP interoperability agreement, that allows ARM and Cadence reciprocal access to relevant IP portfolios.

The reference system includes commonly used functions required in an IoT platform, including a processor, an IoT subsystem, and standard interfaces. It accepts expansion shields for sensors, actuators and wireless connectivity. Utilizing ARM’s MPS2 rapid software and hardware FPGA prototyping platform, ARM and Cadence have built a working system to demonstrate and highlight the benefits of providing IP that has been fully integrated and verified together. The system includes interface IP from Cadence and the IoT processor subsystem from ARM. From Cadence there is MIPI SoundWIre, Quad-SPI, I2C and SPI. The ARM IP includes a Cortex-M processor, power management, flash cache, flash controller and the Cordio radio. The reference system demonstrates that the Cadence and ARM IP blocks are fully interoperable and have been extensively tested. To complement this hardware solution for the creation of an IoT product, ARM mbed provides the necessary software components to build a secure, power-optimized sensor-to-cloud application.

This reference system allows designers to jump-start IoT design projects and to focus more of their design bandwidth on aspects of the design that will differentiate the system in the marketplace. In addition to the usual cost, risk, and time-to-market reasons for using 3rd party IP, the reference system reduces the amount of verification needed since the various IP components have already been tested for interoperability. The time from having an idea to finding out how well it performs in the marketplace is reduced to a minimum.

Using the reference platform a team of just three engineers taped out a design in just three months. So the reference platform helps teams new to IoT to enter the market, and also accelerates getting a design to market for teams already in IoT. Getting to market fast is a double win. Firstly, there are all the usual positives about a short development schedule: lower costs, maybe higher profits. But secondly, one of the biggest issues in IoT is to deliver something and then, as the VCs like to say, pivot. So getting a second version of the design out, once you have a better idea of what the market requires, is just as important.