Internet-connected devices have been predicted to become popular for many years, but the emergence of the internet of things (IoT) has been held back by many issues – the main one being cost.
Making a fridge or other device internet-aware adds significant costs to what is essentially a commodity, low-cost item.
And there are issues around how devices would connect to the internet in the first place.
In many business cases, existing systems, such as production line equipment, were already running on proprietary networks, and within the consumer environment there were competing standards, such as X10 and HomePNA. Wireless technologies also operate on a mix of spectrums, with the 2.4-2.5GHz band being used heavily by several non-compatible standards for data transfer and various remote controls.
Utility companies have long wanted to use the power connections into people’s homes to remotely read their meters. As governments have pushed the need for energy management to lower carbon dioxide emissions, utility companies have also been looking at how they can evolve this into a “smart grid”, where actions can be taken to control certain devices within a property depending on electricity availability or government rules.
The likes of Google with its Google Glass requires the IoT to be more of a reality. Interactions with the world at a technical level in real time will require many more items to be connected to the internet so such wearable technology can understand what is there. There has been a strong move towards a standardised TCP/IP transport mechanism, which has started to erode the technology challenges to a real IoT. Modern PowerLine internet from the likes of Devolo and Solwise can provide networks of up to 500Mbps. Wireless standards are increasing bandwidth regularly, with 802.11n access points from suppliers such as TP-Link and Netgear capable of running at 600Mbps.
And new standards, such as 802.11ad, designed to carry up to 7Gbps, are on the horizon. 4G mobile wireless using Long-term Evolution (LTE) is already being introduced by EE in the UK at speeds of up to 300Mbps. Silicon chips for basic controllers are becoming cheaper, and a device can now be made “internet-aware” for only a few dollars. The capability to switch items on and off or to get them to follow simple commands is now an effective concept. Technologies such as near-field communication (NFC) can be used to transfer small amounts of information, with radio frequency identification (RFID) being a cheap means of enabling items that may need to report data but not receive any.
This convergence, down to a standardised TCP/IP protocol with higher bandwidths available and lower costs across multiple communications, points towards the IoT becoming more mainstream in a short period of time. However, many problems remain that must be dealt with before the IoT can deliver on its promise.
Getting your home online
Let’s take the case of the internet-enabled home. The utility company wants to take control of certain items, such as the freezer. It can save energy here by turning the freezer off for periods of time, as long as it knows the temperature within the freezer, and turns it back on again before the temperature rises too far. You might like to control the cooker and other devices via the internet, so when you are on your way home, a nice meal will be ready as you walk through the door to a house at just the right temperature, with music playing in a room lit at an ambience to suit your mood. The TV company wants to monitor what you watch, so it can deliver highly targeted advertising to you in a mutually agreed manner. Your local supermarket wants to monitor your fridge and freezer to see what has been taken out and not returned so it can build up a shopping list for you.
There are some feasibility aspects to consider though. First, should each party connect through a different medium? If we want to ensure the connected property operates across a range of interested parties, it is important that everything works together. Therefore, a single means of connectivity will be required, which for most will be an existing broadband connection. No need for utility PowerLine connections or other semi-proprietary systems. Just use a single connection to the home. Increasingly, homeowners may want a level of redundancy and so have more than one load balanced and redundant connection, but essentially, all communication to do with the IoT should be across a single point of control.
Second, let’s look at what could be connected: heating, kitchen and laundry equipment, entertainment systems, and other mainstream connected devices. Many of these will require constant monitoring. For example, in the case of the freezer the utility company has just turned off to save energy. It may seem sensible to check every half an hour to see if the freezer needs turning on again, but what happens if the house owner opens the fridge door and leaves it open for a few minutes?
Many IoT devices are capable of large amounts of low-level data “chatter”. Each item of data may be small, but the quantity of them will make for a large amount of noise. For a utility company monitoring several million households, the bandwidth required to carry this out will be high.
Therefore, something needs to be done to control this chatter. An intelligent property should be able to minimise the chatter. Imagine a property where all the devices are automated through simple internet enablement. Rather than have each of these act in isolation, a meshed network of a more intelligent set of controllers – such as a network of Raspberry Pi microcomputers – could deal with the chatter. For example, as long as the freezer is within its temperature bounds, the utility company does not need to know about it. The Pi network can filter unwanted data close to its source.
Layering over suitable data analytics, from the likes of Acunu or Splunk, could add further value in being able to monitor trends and carry out pattern matching for activities where actions should be taken.
An intelligent internal network can also lower the amount of external data required to be pushed into a property for actions to be taken. Again, taking the previous example of someone returning from work and wanting a set of actions to be taken by several devices, rather than sending a command to each device separately, a command can be sent to the network saying something like “initiate environment A”. The internal network will understand this, parse it down to a set of commands and make each one happen. What could have been 20 different commands across the public internet has been brought down to one command.
The same approach can be used in business, with single commands being passed along value chains and kicking off complex activities within specific networks. Communities can work in the same way – centralised receiver systems could accept single requests from citizens and split them into more complex sets of events. For example, a council house tenant may complain that their water is too hot, yet the house is too cold. The council can send out a single command that turns down the temperature of the water while increasing the thermostat temperature in specific rooms. The key is to control how the IoT talks at a meaningful level. No matter how small the packets of data being sent out and received by an individual item are, if these are across billions of items, the aggregate data will bring the internet to a standstill. Through the use of localised intelligence, IoT chatter can be minimised.