IoT and Energy Consumption: Will Connected Devices Exhaust Our Power Supplies?
In recent years, the rapid propagation of connectivity has fueled a race to develop smart devices, from the development of smart homes through to automobile functionality. This hype has its downsides, however, clouding the reality of how many IoT devices will be operational in the coming years.
As such, the timescale and rate of growth for these devices differs wildly between various research and auditing companies - KPMG forecasts that by 2020 there will be upwards of 20 billion IoT connected devices, yet Intel raises that to 200 billion in the same time frame.
Whatever the eventual number of devices, we can all agree that it is a lot. The ongoing implementation of IoT infrastructure will drive a fundamental shift in energy consumption and management. On the one hand, connected devices and their capabilities in monitoring and analysis of machinery allow for ambitious energy saving techniques, at both a consumer and industrial level. On the other hand, the sheer number of devices running on wireless networks, added to the data centres they are connected to, poses an issue of theoretically vast energy consumption.
BDJ spoke with two IoT experts to get their insights into the changing nature of IoT energy consumption - Andy Stanford-Clark, CTO at IBM for the UK and Ireland, an IBM Master Inventor and Distinguished Engineer for the IBM Watson Internet of Things Platform, and Stacey Higginbotham, former Senior Editor at Fortune and creator of the influential IoT Podcast.
High Energy Requirements
An increase in IoT devices and larger networks to accommodate them will naturally produce a vast amount of data that will need to be transmitted and stored. A 2016 study determined that without advances in efficiency, the combined ICT industries could consume 20% of all the world’s electricity by 2025, and be responsible for up to 5.5% of all carbon emissions.
The question is, can the established infrastructure cope with this energy demand? The increase comes at a time when economic, political and social movements are focusing on emissions targets and the drive towards renewable sources of energy.
“The infrastructure is already evolving to deal with the conflicting demands of more devices and the need to use less power,” Andy explains. “The power grid is rapidly adopting new technologies for load balancing, load shedding, peak shaving, demand response, and power storage technologies that help to balance out the intermittency of renewable generation against the demands we place on the power grid as consumers.
“The switch to renewable will actually lead to the demand for more connectivity”
“Managing the charging of electric vehicles, for example, will be essential to ensure we don't overload the power network when we all come home from work, plug our EV to charge, and turn on the cooker to make dinner at 6pm each evening.”
Stacey also sees the shift away from fossil fuels for energy generation as beneficial for the overall adoption of increased IoT connectivity. “The switch to renewable will actually lead to the demand for more connectivity,” she says. “This is because that will help us get an accurate projection of demand and let us anticipate what we need on normal days.
“When catastrophes happen, we'll also be better able to track usage and figure out what it's from and shut off non-essential uses. Additionally, the IoT should help prevent unplanned equipment failures in utilities that could help reduce emergencies.”
Of course, a global shift towards renewable sources of energy has not been realised on a comprehensive scale as of yet. The rate of IoT device propagation, though, will not likely slow to accommodate a change in energy generation methods. How, then, can the IoT achieve realistic handling of its logistical demands in this period when the fundamental aspects of energy generation are changing?
The answer, according to Stacey, is in the software: “We should focus on building energy-efficient software, focusing on energy harvesting technology for sensors and the overall optimisation of data center gear to cut energy demand.”
How IoT Could Save Energy
While the growth of wireless network-enabled smart devices is set to skyrocket, thereby increasing energy demands, there is scope to utilise the analysable nature of these devices in order to save energy.
“What we'll start to see more of is devices powered by ambient energy harvesting - extracting power from things like radio waves in the air around us”
“Most IoT devices use less than 1 Watt of power when they're ‘on standby’,” Andy says. “What we'll start to see more of, though, is devices powered by ambient energy harvesting - extracting power from things like radio waves in the air around us; vibration from machinery, vehicles, etc; heat and solar energy, etc.
“Therefore, over time, devices will become effectively self-powered and we won't have to plug them into the mains or change the batteries every few months. Lower power radio communications systems like BLE (Bluetooth Low Energy) and LoRa ( low power long range ) are helping this drive, but reducing the power demands of getting data from the sensors to the Internet.”
The potential for IoT connectivity to improve efficiency of energy-intensive processes is far reaching, spanning various industries. One example gaining widespread attention is from an industry that remains illegal in many parts of the world - the cannabis industry. The energy-intensive nature of cultivating cannabis plants has long been a large overhead for professional growers, but with the advent of IoT-enabled sensors, there is the opportunity to vastly improve the efficiency of maintaining optimal greenhouse conditions.
Rather than relying on manual adjustments to heat and water outputs, both large and small scale cannabis cultivation operations are installing said IoT sensors to assess growing conditions, thereby reducing the wastage of raw materials. These sensors can also be used to gather data for the purposes of deeper analysis, in order to determine the presence of microclimates inside the greenhouse that may affect longer-term cannabis yields.
Digital twins are also proving important. These are software models of physical systems, on which you can run ‘what-if’ scenarios to find optimised operating models, and Andy explains we are seeing them used more and more in order “to compute machine learning models to run all kinds of equipment more efficiently, increase the operational lifetime, and reduce maintenance costs of hardware. Examples include iron smelters, industrial processes, vehicle, train and shipping movements.”
Proven Industry Examples
Various industries have proven themselves more than capable of facilitating energy efficiency in their processes. In an IEEE Spectrum article from September 2018, Stacey highlighted the estimated $60 billion saving per year that companies like Google and Amazon had achieved since the early 2000s by prioritising the metric of performance per watt, and forcing their suppliers, including Intel, to do the same.
Stacey also uses the example of the electrical component manufacturer Schneider Electric, and their factory in Lexington, Kentucky. By incorporating sensors into their operations, they have been able to engage in detailed analytics of the manufacturing lines, focusing on the mix of products being made, and in which order. Stacey details the positive results:
“After tweaking the production mix, the plant reduced consumption by 12% in year three and 10% in year four. ‘The entire group had been focused on the processing side, and now every process decision is dictated by energy savings,’ says Andy Bennett, former senior vice president of Schneider Electric’s EcoStruxure platform, which drove the factory innovations. ‘What has changed in the last five years is the technology and the drive and need to have a sustainable message.’”
As the above examples highlight, the use of the analytical power of IoT sensors and data centres for the purposes of energy optimisation is already a reality, and could be extended across any number of industries.
Planned Obsolescence and the E-waste of IoT
However, the increase in connected devices does present another potential problem for the environmental impact of the IoT. “The challenge is more about e-waste and longevity of those batteries,” Stacey says.
The UN released a report in 2017, detailing the scale of e-waste around the globe, and the factors contributing to its increase. In 2016, an estimated 44.7 million metric tonnes were produced globally, working out at 6.1kg per inhabitant. This is expected to rise to 52.2 million metric tonnes by 2025, or 6.8kg per inhabitant.
A number of factors are contributing to this rise in e-waste. The number of mobile cellular and broadband networks has increased around the world, allowing more people access to the internet and therefore increasing the demand for mobile phones, tablets and other connected devices. The prices of these electrical items have also fallen, resulting in consumers buying more than one device. The lifecycle of a smartphone has fallen too, with an underlying culture of frequent upgrades and forced obsolescence.
The growth of the IoT has gone hand-in-hand with the placing of semiconductors in traditionally ‘dumb’ items. While many instances of this have offered clear advantages and consumer uses, e.g. smart watches and fridges, there have also been examples of more superfluous uses of the technology.
For example, Wilson sells a basketball with an inbuilt sensor and bluetooth connectivity that tracks users’ shots and activity. If the sensors break, the logistics of opening up the basketball to replace them are far less feasible than opening the back of a smartwatch to replace a battery. Extrapolate this out and items like this basketball are an indication of the potential scale of the IoT-fuelled e-waste problem in years to come.
Is IoT Sustainable?
A complex symbiosis is apparent when examining the deployment of IoT-connected devices and their effect on subsequent large-scale energy requirements. Intelligent deployment of sensors at a manufacturing and industrial level has a proven track record of waste reduction through process optimisation, thereby offsetting the inescapable nature of more smart devices on the market requiring more energy to run.
More of these sensors and connected devices will be needed in various levels of industry and manufacturing in order to conduct the necessary analysis that creates actionable optimisation that saves energy.
“If something is invented which literally *everyone* feels they have to own, then that will push the numbers up”
Andy explains how the continued development of IoT devices also has the potential to fundamentally alter how raw data is processed, especially by introducing machine learning algorithms to increase efficiency:
“Reducing communications by putting more intelligence at the ‘edge’ of the network (in the devices themselves) is becoming more prevalent. Rather than streaming lots of sensor data to the Cloud for processing (and throwing most of it away), the key here is to distribute the processing of the raw sensor data into the network and down to the devices, so they send back ‘information’ rather than data, i.e. useful stuff distilled from the sea of raw data.
“Training machine learning models will still need to be done in the Cloud, as it requires heavy computing power and the benefit of large amounts of historical data from many devices. But once you've got a trained model, the run-time requirements for that are quite modest, and could even be burnt into silicon in a low-powered chip in the sensor device.”
The levels to which the optimisation of energy expenditure can be achieved, especially with the ongoing shift towards more renewable sources of energy, will be the crucial element in determining the IoT’s long-term environmental impact. Predictions for the number of IoT devices in circulation within the next 5-10 years vary wildly, which makes a true blueprint of sustainability almost impossible.
“It is difficult to make accurate predictions, especially about the future,” Andy notes. “The problem is that we have no idea what the ‘killer apps’ will be for IoT. If something is invented which literally *everyone* feels they have to own, then that will push the numbers up. If it's something that is on every product we buy, that will contribute a load more.”
As the Internet of Things has developed, as have the myriad ‘smart’ products riding the wave of hype around it. Smart fridges and smart watches have been the typical examples used when people explain the consumer use cases for IoT technology, but not every solution is so well received. Products that broadly fall under the umbrella range from the fanciful to the outright ridiculous, with connectivity for connectivity’s sake affecting everything from hairbrushes to condoms.
One example is HidrateSpark, a water bottle that tracks your water intake and glows to remind you to drink it. It raised over $600,000, but has been largely ridiculed. A water bottle simply does not have to be smart - not having to remember to drink water is the product’s main selling point, something you should probably do anyway. Egg minder will let you know when you’re running low on eggs and even how fresh your eggs are, a groundbreaking replacement for checking the box for the quantity and expiry dates. Trakz is a fitness tracker for your pet, which can tell you how much your pet is eating - the less said about this one the better.
Illustrations by Kseniya Forbender
To contact the editor responsible for this story:
Margarita Khartanovich at [email protected]
- How Blockchain Can Reshape Charitable Donations
- Blockchain’s Scaling Crises: Can Sidechains Be A Potential Solution?
- Hacking Blockchain: Is it Really Secure?
- Regional Strengths Are Shaping AI’s Evolution in Asia
- Credit Card vs. Bitcoin: How Do You Pay for Your Coffee?
- Do You Trust AI? This Is What You Must Understand to Do So