Internet of Things

An IOT device is often just the starting point in a connected architecture of cloud storage, analysis and visualization. Processing done on the device is critical to the correct functioning of the entire system.

IOT devices are small circuits built around microcontroller chips. The circuits add sensors, memory, wireless connectivity, battery monitoring and other functions to the controller. In order to end up with a finished device, the microcontroller needs to have program code that links everything together in a cohesive way.

Building good IOT code requires a very different set of skills from the norm. JTA has physicists, electronics engineers and specialist programmers ready to take up the challenge.

In the Internet of Things area, programming microcontroller chips (“MCUs”) happens at a very low level . They can have an operating system running on the chip but mostly they don’t to save space in their memory. Programming is usually done using the C language and the final code moved over to the MCU for testing. A simulator program may also be used instead of hardware.

Programming an MCU is an art. The systems don’t have much memory. Also every operation will use energy and drain the battery. Consequently, it is vital that the programmer understands the impact of every instruction on the hardware of the chip and on the extended circuit.

In this Internet of Things scenario, to help with battery life there are many useful techniques that the programmer can deploy on the chip. Examples include altering the speed at which the chip processes instructions. While simple, it is possible to save a lot of energy this way. The trick is to find the sweet spot where the processor is running fast enough for the job at hand but no more.

There are also methods by which the processor can switch parts of its brain on and off to save energy. A common technique is to set up a kind of alarm call and let the processor sleep. When the alarm call activates, the processor will do the task required and can then go back to sleep. This is much more efficient than having the chip always active.

Information storage is also very important. When we write programs in a so-called, high-level language the language will have a set way to convert common items like text and numbers into bits and bytes. When we work with an MCU the programmer can develop their own internal representations. This is harder work but it gives us the opportunity to design an internal representation that is more efficient or could significantly assist with subsequent calculations. Lastly, the internal representation can employ compression and decompression techniques to save memory, and, often, more importantly, the use of the wireless systems to upload data to the cloud.

• Consider the way in which data will be stored so that the size requirements are minimized and other processes made more efficient
• Consider battery usage and employ techniques to shut down the processor and other peripherals when not needed
• Build in data compression techniques to save space and Bluetooth power
• The programming team should be involved in the hardware design. Their code will often have a very close connection to the hardware in order to be able to control it well
• Build a prototyping board to host the proposed circuit but with test points to be able to attach monitors to measure how the circuit responds to programming changes and different power saving techniques
• Build input simulators to generate random, but representative, input data for extended test runs
• Log all of the vital signs from the prototype so that we can employ data science techniques to monitor the system for unexpected behaviour