The growing proliferation of next-generation technologies such as internet of things (IoT), 5G and artificial intelligence (AI) has presented new energy management challenge for enterprises. While on the one hand technologies such as smart meters are being harnessed by companies to optimise their energy consumption, on the other, processing huge volumes of data generated by these devices is driving the need to deploy more data centres. Most data centres consume large amounts of electricity, thereby resulting in high energy costs for businesses. The jury is still out on the net impact of the growing adoption of advanced technologies on the energy footprint of enterprises. However, given the fact that the uptake of these technologies is only set to grow exponentially in the years to come, the need is to devise energy efficient strategies to reduce the overall power requirements. tele.net takes a look at the energy needs of select next-generation technologies and the associated energy management strategies…
IoT
IoT is being steadily embraced by enterprises across every industry to streamline their business operations, achieve greater time and cost savings and deliver new value propositions to customers. In the next few years, IoT is expected to take most of the human-driven processes beyond their conventional realms by integrating advanced technologies such as analytics, cloud computing, AI and robotics. Even though the average energy consumption of a single IoT device is minimal, the humongous amount of connected devices that are likely to be deployed in the next few years raises concerns about the total energy requirements of IoT networks. The number of internet-connected devices (12.5 billion) surpassed the world population (7 billion) way back in 2011; they are expected to reach around 20 billion globally by 2020.
Industry experts, however, are of the view that IoT devices will become effectively self-powered over time, drawing energy from the environment including radio waves, vibrations from machineries and vehicles; heat and solar energy. This will reduce the replacement cycle of the batteries. The growing popularity of low power radio communications systems such as bluetooth low energy (BLE), low power wide area network (LPWAN) and long range (LoRa) is expected to further drive this trend. LPWAN is particularly designed for sensors and applications that need to send small amounts of data over long distances at short hourly intervals every day. Meanwhile, LoRA is a non-cellular version of LPWA, which is available in licence-free spectrum and offers low power and low data rate communication over long distances, enabling battery-operated devices to function for up to 10 years without any human intervention.
A more formidable challenge, experts contend, is that of managing the e-waste generated from IoT devices. According to a 2017 United Nations report, an estimated 44.7 million metric tonnes (mmt) of e-waste was generated globally in 2016. This is expected to rise to 52.2 mmt by 2025 and IoT will be one of the key contributing factors to this increase, primarily on account of the large-scale use of batteries and semiconductors in the connected devices. Concerns are also being raised about the superfluous uses of the IoT technology such as placing sensors and bluetooth connectivity inside basketballs to track 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. This points towards the potential scale of the IoT-fuelled e-waste problem in the years to come.
5G
Telecom operators globally are gearing up for the commercial launch of 5G services. 5G promises to deliver ultra-high speeds and latencies of as low as 1 millisecond, compared to 4G, where latency is 50 milliseconds. 5G is, therefore, more than just an incremental technology to 4G and its deployment will require massive technical and infrastructural upgradation of the core and radio networks and spectrum. These transformations will be highly complex as well as capital and energy intensive in nature. However, there are concerns that if 5G promises to offer much greater speed compared to 4G, a similar rise in energy consumption could follow, assuming energy efficiency is kept constant.
Experts attribute this to two elements, which are expected to be fundamental parts of 5G networks, such as 5G small cells and massive multiple-input multiple-output (MIMO) antennas. It is estimated that 5G small-cell deployments will overtake 4G small cells by 2024, with the total installed base of 5G or multimode small cells in 2025 to be 13.1 million, constituting more than one-third of the total small cells in use. Although energy consumption in a small cell is much lower than that in a conventional cell, many more small cells are needed to cover a given area, thereby increasing the total energy consumption of the network.
Meanwhile, massive MIMO technology involves the use of arrays with several antennas at each base station, thereby increasing the total energy consumption of 5G base stations compared to 4G. Industry experts, however, are optimistic that as massive MIMO technology develops, its energy efficiency may also improve over time, and despite containing more hardware, future massive MIMO base stations will consume less energy than 4G base stations. Moreover, future improvements in massive MIMO hardware are also expected to serve many more users at the same time and frequency, thereby reducing the average energy consumption per user.
Another reason that will probably make 5G networks less energy intensive compared to 4G networks is the fact that 5G base stations can be put into a hibernation mode whenever there are no active users. In contrast, 4G networks need to continuously transmit a lot of control signals even when there are no active users, for example, at night. Thus, 5G networks are potentially capable of putting more components to sleep for a longer time, thereby reducing energy consumption by almost 10 times compared to current systems when there are no users.
AI
AI is being increasingly explored by leading global operators to generate cognitive insights by analysing huge volumes of data collected by them. It is also being used to automate back-end business processes, which frees up staff, thereby increasing productivity and lowering costs. While AI has the potential to reduce power requirements of enterprises, in the current stage of the technology, training AI-powered machines is turning out to be highly energy intensive and time-consuming. A recent report suggests that the carbon footprint of training a single AI-powered car is as much as 284 tonnes of carbon dioxide equivalent – five times the lifetime emissions of an average car. AI-enabled machines are trained via deep learning, which involves processing vast amounts of data and, therefore, significantly increasing computing and, thereby, power requirements. For instance, an AI-powered chatbot is fed billions of written articles so that it learns to understand the meaning of words and how sentences are constructed. To address this issue, major technology firms such as Amazon and Google have started offering cloud-based platforms that researchers can pay to use remotely for training AIs. However, for the next few years, till the technology is commercialised, energy requirements associated with AI-enabled machines are poised to grow significantly.
The way forward
Energy requirements of the information and communications technology (ICT) sector have increased substantially in the past few years, owing to the deployment of connected devices and the concomitant increase in the demand for computing infrastructure to process huge volumes of data generated by these devices. A recent report suggests that ICT industries globally could consume 20 per cent of the entire world’s electricity by 2025, and will be responsible for up to 5.5 per cent of all carbon emissions. Going forward, the need is to develop an energy infrastructure that meets conflicting demands of more devices and the need to use less power. Efforts must be made to reduce the energy consumption of data centres by replacing older servers with newer and more energy efficient units, improving the efficiency of cooling by exploiting outside air or water sources and, as far as possible, meeting the energy requirements through renewable sources. Moreover, there is a need to design energy efficient sensors for smart devices and improve computer processing capabilities to process big data more efficiently and with minimum power consumption.
