Smart Dust

In the mid-90s, those interested in wireless sensor networks began talking about ‘Smart Dust’. They envisaged that it would soon be possible to create incredibly small, wirelessly connected sensing nodes. Not unreasonably, they expected that these tiny sensors would have a myriad of life changing applications.

Smart Dust would be so small and so cheap that you could literally sprinkle it over surfaces, embed it into paint or other materials, inject it into people, and so on. Then, once the dust had settled (so to speak) it would start communicating sensor readings and a whole host of amazing ‘smart systems’ would be possible. Your house might be able to monitor your position at all times, and react accordingly. You could paint smart dust over industrial equipment and monitor its condition in incredible detail. Smart Dust in museums could monitor storage conditions, help an alarm system and monitor people flow all, at the same time.

These dreams of ‘smart’ systems and buildings, reacting to your every action, have been commonplace for years. But despite the different parts of the technology slowly reaching maturity, the whole picture seems to continually elude us. So where is our Smart Dust, and how can those of us in technology development conjure it into existence?

The problem with dreaming big is that it often involves solving problems across many different domains. In order to realise an application of Smart Dust in which your house and clothing can track movements and control objects, we have to solve many tricky challenges:

  • Size
  • Connectivity
  • Data Processing/User Experience
  • Interoperability
  • Security

The number of engineering domains involved is surprisingly large. The system as a whole involves everything from engineering a front end user interface for the house owner, through to building an antenna the size of a speck of dust  and which can be usefully used to communicate with another node.

The degree to which all the problems are coupled is also unhelpful. Even at larger sizes, for instance a few cubic centimetres, the amount of available power becomes so problematic that it can affect almost every part of the system. The small battery size restricts radio communication, which in turn affects how much data you can send, which in turn restricts what the application can do. At least with today’s technology outlook, we shouldn’t expect streaming video over ultra-low power sensor nodes. At the same time, these types of restrictions cover more than one problem domain, causing hidden couplings between the design choices which could make the total product difficult to design. For instance, unless the hardware engineer is talking to the application engineer, there’s a possibility that the battery life will be inadequate. Few companies have the required breadth of engineering expertise under one roof.

So physically building Smart Dust is extremely difficult. It requires state-of-the-art component miniaturisation and the development of protocols for the network that require tiny sips of radio time for power efficiency, but can still perform a useful amount of routing and forwarding. Today, many systems avoid the networking problem by having mains powered routing nodes - which have effectively unlimited life - and then allowing the sensors to sleep indefinitely, waking when required. This creates a need for fixed infrastructure and vastly reduces the range of possible applications.

So far so negative, but you may be surprised to learn that much of the Smart Dust dream is already a reality. The dust itself doesn’t yet exist, but that hasn’t impeded the development of ‘smart’ systems.  Today I can easily buy an automated heating system which will learn my routine, and react to my demands. My phone automatically gathers information from a ‘smart’ pedometer I carry and uploads the data to the cloud, where information about my activity can be calculated. But these systems all require charging, managing, and a bucket load of human interaction. They’re not quite the dream of the building that simply knows where you are and what you’re doing, but they are a passable imitation.

The core challenge right now concerns interoperability. Today, it’s still impossible for me to let my ‘smart’ heating system use the sensors from the wireless outdoor thermometer that I bought separately. If I purchase a smart fire alarm from one company, I cannot use it with a different system. My games console can recognise my face, but I still have to login to the television catch up services separately. This interoperability issue, wrapped up in security and commercial complexity, may ultimately be overcome but progress is slow.

A related issue is that sensors are normally built specifically for a particular service, and then it’s only the services that share information. This presents a huge vertical technology challenge, as sensor design is influenced and narrowed by the service. Imagine instead a future where sensors are interoperable and where a single gateway can connect many types of sensors, all using similar wireless technologies. In this system, the user owns a pool of sensors, and decides which services can know what. Services can access more information and users can choose the right sensors to fit their lifestyle and use case.

It’s not all doom and gloom though: A version of interoperability is being enabled by a ‘straight to the cloud’ approach. Rather than devices talking directly to each other, they communicate back to the cloud, where interfaces can be provided regardless of the physical constraints of the protocols being used in the home. This is the way that some sports and fitness services have gained traction, achieving a form of interoperability through app-to-app and cloud-to-cloud sharing – Apple’s HealthKit being the most notable example. However the smartphone is not an ideal enabler for smart home applications, as people tend to take their phones with them when they go out. Devices such as Amazon’s Echo are a neat way around this problem, but tend to remain locked in their own corporate ecosystem.

Perhaps then Smart Dust won’t be as small as we originally thought. Picture instead a collection of different sensors in a variety of shapes and sizes, working together seamlessly. Rather than carry a specific smart pedometer with me, I could simply wear my smart enabled shoes. Instead of having a 'smart thermostat with its specific boiler box, my boiler might simply monitor readings directly from sensors in the house.

Interoperability and standardisation are very difficult indeed. Few vendors wish for their systems to work with other, wider systems. Some companies don’t want you to understand how their systems function at all. If and when all sensors can talk to each other there will be a pressing need to control who can see all of that data, and what they can do with it. I’ll want to make sure that burglars can’t simply ask my house when I’ll be out, and schedule to come back later…

So where is my Smart Dust? Well, like the jet pack and flying cars, it’s taking its time and may never arrive. But something, a version, really is coming. In the short term, the larger systems that exist are beginning to prove the promise of smart systems, albeit locked in their little walled gardens. But this will only be the proving ground for the technology. Eventually the technical hurdles surrounding ad hoc mobile networking will be overcome, allowing us to have true meshes of many different sensors. When that happens, we'll break out of the walled gardens with a much better idea of where we're going and a set of compelling services, ready to make use of the rich content that Smart Dust can provide.

Dr Dominic Crutchley