in English

Alternative Solutions for powering IoT devices

Power availability is a big issue for edge devices in many IoT applications – not so much for actuators, but very often for sensors. Consider, for example, a tracker application in which a device is dropped into a parcel consigned for shipment; the user can track the parcel during its journey and check its progress or even if it has been mistreated. Clearly, no wired connections are possible, so the device must communicate wirelessly while relying on an internal power source.

In other scenarios, edge devices may be located in remote areas, and possibly in large quantities. No mains power is available, while maintenance visits to replace batteries are time-consuming and expensive.

Edge sensor designers can respond by using rechargeable batteries or supercapacitors together with an energy harvesting strategy to generate charging current for them. Energy harvesting is attractive, as it taps into an inexhaustible supply of ambient energy, but it is also challenging as it may not meet the power needs of the node without careful design.

Therefore, taking every possible step to minimise the node’s power demand is essential in achieving viability for the energy harvesting approach. Even if a standalone battery must be used, power optimisation remains important. Saving just a few micro Joules a second can mean changing a battery only after 10 years instead of every year.

In this article, we look at why edge devices need power, and the options for reducing this need to a level where energy harvesting becomes sustainable. Next, we examine some examples of the most popular energy harvesting technologies available, and see how these can be complemented by suitable storage devices.

Identifying and mitigating edge device power-demand culprits

Edge sensors’ power requirement arises primarily because a sensor device must digitise, package and transmit any measured data to be useful. This typically calls for at least the sensor, a microprocessor, complete with crystal oscillator, memory array and possibly A/D converter, and the communications interface hardware.

A crucial approach to minimising power demand is to choose a processor that can be efficiently driven into sleep mode when no measurements are necessary, and draws minimal leakage current while asleep. Also essential is the right choice of local network protocol; some protocols may have more data bandwidth than required, while drawing excessive power to support it.

Other techniques for energy reduction are also available; these relate to the choice of processor, memory subsystem, oscillator and A/D converter, as well as coding efficiency.

In many cases, energy harvesting solutions can be used to deliver the power required; either because demand has been successfully reduced using the approaches above, or because ambient energy is available in adequate quantities. We review some of these below, together with some storage technologies that can be used to complement them.

Energy harvesting technologies

There are four main ambient energy sources available in the environment: mechanical, thermal, radiant and biochemical. These energy sources are characterised by different power densities. The most popular are radiant and mechanical, as illustrated by the examples that follow.

Solar energy

Solar energy harvesting is facilitated by photovoltaic (PV) cells that convert sunlight into a flow of electrons due to a photovoltaic effect. Solar cell technologies have traditionally been divided into three generations : First generation types are mainly based on silicon wafers and typically perform at about 15 – 20% efficiency. They offer good performance, but are rigid, and their production is energy-intensive. Second generation cells are based on amorphous silicon and deliver typically 10 – 15% efficiency, and have some flexibility. Production costs less than for first generation, but is still energy intensive. Additionally, use of scarce elements limits price reductions.

Third generation solar cells use organic materials such as small molecules or polymers. While their performance and stability is currently limited compared to first and second generation products, they offer great potential for advantageous pricing and efficiency, and are now being commercialised.

Fig.1: Solar energy harvesting is facilitated by photovoltaic (PV) cells – Image via Wikimedia

Solar power can allow IoT devices to be powered indefinitely, but there are challenges in applying it, especially if the devices are small. Many solar technologies generate products that are too bulky, rigid or inefficient for use in small and remote IoT devices; one exception is Alta Devices , which offers Gallium Arsenide (GaAs) solar cells that are light weight, flexible and thin – so much so that they can be designed into specific products by moulding them around curved surfaces.

These cells operate with a world record efficiency of 28.8%, which permits greater power from a smaller surface area; a property which can enable more IoT sensor applications with critical power demands. In particular, they can harvest useful power levels from indoor or artificial light, which typically does not produce the full spectrum present in sunlight. For example, in the dim lighting conditions of a warehouse, where light levels could be 200 lux, a single Alta Devices cell could still generate hundreds of microwatts – sufficient to power an IoT sensor if used with a suitable form of energy storage.

Solar panels are used to power larger IoT devices as well. Wireless parking meters from IPS Group Inc. are highly power-efficient and harvest energy through a miniaturized PV solar panel and store it in industrial grade TLI Series rechargeable Li-ion batteries. These autonomous meters connect to the IIoT through a wireless network to permit more reliable and efficient billing and reporting. They also reduce pollution by alerting nearby motorists when a parking space becomes available.

Fig.2: Solar-powered parking meters – Image via Flickr

Compact solar panels can also be used in applications that are extremely environmentally demanding; e.g. a solar-powered CattleWatch ‘smart collar’ which is exposed to mechanical stress as well as liquid ingress.

These collars allow a herd of cattle to form a wireless mesh network. Ranchers can remotely manage the herd through connectivity via Iridium satellites to the IIoT. Industrial grade TLI Series rechargeable Li-ion batteries enable the ‘smart collars’ to be lighter and more compact, and thus more comfortable for cows to wear.

Mechanical or kinetic energy harvesting

Kinetic energy from everyday activities can potentially be used to power smart devices. University of Wisconsin-Madison engineering researchers Tom Krupenkin and J. Ashley Taylor have developed an in-shoe system that harvests the energy generated by walking; they claim that up to 20 W of energy could be generated, and stored in an incorporated rechargeable battery.

The technology is based on a proprietary process known as ‘reverse electrowetting’ which converts mechanical energy to electricity via a microfluidic device, in which thousands of moving microdroplets interact with "a groundbreaking nanostructured substrate". The process acts with various mechanical forces, and can output a wide range of currents and voltages. Ways to link the battery to the phone using conductive textiles and wireless inductive coupling are also being investigated.

Fig. 3: EnOcean push button multi-channel switch module

Very low levels of kinetic energy can also be used in building automation and control systems; for example, EnOcean’s pushbutton modules incorporate rocker switches that operate on an actuating force of 7 N over a travel of 1.8 mm. Up to four pushbuttons can be accommodated within a single wall-mounting module. An energy converter, which works like a dynamo, converts the switch action into electrical energy with an output power of 120 µW. A radio signal is generated when each pushbutton is either pressed or released; this contains a button code and unique module ID.

These pushbutton modules are complete edge sensor devices that manage the signal all the way through from the switch to a reliable, energy saving radio protocol. This not only ensures that the low level of kinetic energy is successfully harvested for the application, but also that the product is easy for systems developers to integrate into their building automation, home automation or other project.

RF energy harvesting

RF energy can be used to trickle charge or operate consumer electronics such as e-book readers or headsets, wearable medical sensors and other devices. This power source has plenty of potential because of the large and growing number of radio transmitters around the world, related to TV and radio broadcasting as well as mobile phones and other devices.

Fig. 4: Pet implant RFID chip; the rice grain indicates size – Image via Wikimedia

Radio-frequency identification (RFID) is a special form of RF energy harvesting, in which power, as well as information, is obtained from a specific source rather than random ambient RF energy. An RFID base station has a scanning antenna which transmits radio-frequency signals over a relatively short range. This communicates with a transponder built into a passive RFID tag – and provides the transponder with the energy necessary to wake up, read and communicate in response. There are active RFID tags with batteries that operate at a greater distance, but passive tags without batteries have a virtually unlimited lifespan.

Applications are widely diverse: examples include credit cards, real time location systems (RTLSs) to track worker movements or the effectiveness of a store floor plan, asset tracking, motorway tolls and pet ID implants.

Wind energy harvesting

Some techniques allow wind energy to be converted into vibration energy for harvesting. This has been demonstrated by a research project conducted at Chongqing University , China, to build a wind-powered temperature sensor node. The complete device measured 62 x 19.6 x 10 mm and comprised of a temperature sensor, a piezoelectric wind energy harvester, a microcontroller, a power management circuit and a radio transmitter module. The critical wind speed proved to be about 5.4 m/s. If this increased to 11.2 m/s, the device provided 1.59 W into a 20 kΩ electrical load. This was sufficient power for the wireless sensor node to measure and transmit the temperature every 13 s.

Fig. 5: Chongqing University wind energy harvester prototype – Image via MDPI

Supercapacitors and batteries

In many harvesting applications, the ambient energy source may be insufficient to support the sensor, with its processing and communications requirements, directly. In others, the energy may not be available when it’s needed – no solar energy at night, for example. However, these problems can usually be overcome because IoT sensors typically do not have to gather data continuously. If so, a viable solution can be based on sensor electronics that sleep most of the time, only awakening to process a burst of data from the sensor. Meanwhile the harvesting device accumulates energy continuously, or at least whenever the ambient supply is available. The success of such schemes depends on providing a storage medium suitable for collecting the energy harvesting device’s output. Either supercapacitors or rechargeable batteries can fulfil this role. The choice between these depends on the application, as both have their advantages and disadvantages.

A supercapacitor, or ultracapacitor, is a high-capacity device with capacitance values much higher – but lower voltage limits - than other capacitors. It typically stores 10 to 100 times more energy per unit volume or mass than electrolytic capacitors, can accept and deliver charge much faster than batteries, and tolerate many more charge and discharge cycles than rechargeable batteries. It is light weight and low cost. However it has low energy density, low voltage output from each cell, and requires sophisticated electronic control and switching equipment. It can also be prone to excessive self-discharging, wasting much of the harvested energy.

If a rechargeable battery storage solution is preferred, lithium-ion is a popular choice for IoT applications. Other options, such as lithium-ion polymer (li-polymer) are available, and there is always the promise – or hope – of groundbreaking new technologies. Although much research continues around the world, no real game-changers have emerged. This is probably due to the technical and commercial difficulties of converting a laboratory prototype into an economically viable product that can ship in volume and perform reliably and safely. Also, manufacturers are committed to current production techniques and mineral supply deals that would be hard to break.

However, there is optimism for change; for instance, Dyson has invested $15m into Sakti3, a company whose solid-state battery technology promises cheaper, safer, more reliable, more energy-dense and longer-lasting products. Meanwhile, better performance can be obtained by choosing existing batteries that use an improved version of the existing lithium-ion chemistry. Tadiran industrial-grade batteries, for example, offer 20 years’ operating life in extreme environmental conditions while reducing many of the problems associated with standard consumer products. Up to 5000 life cycles are possible, together with wider operating and charging temperatures, high current pulse ability, low annual self-discharge rate and a low leak risk.

Conclusion

The IoT is providing deeper insights than ever previously available into the world around us. New possibilities for information and control range from understanding wildlife behaviour in remote forests to improving driver experience and reducing traffic congestion in large cities, or allowing manufacturers to better understand and improve their production efficiencies. The IoT is just at hard at work in large infrastructure projects as it is in more immediately apparent applications such as home automation and wearable devices.

This success depends heavily on arrays of edge sensors, often in large numbers and frequently over a wide geographical area. Under these circumstances, edge nodes that are low cost, compact, rugged, reliable and very low-maintenance become essential. Using energy harvesting whenever possible helps to achieve these objectives, by making the devices battery-free, or at least run from batteries that will last for many years before needing replacement. Maintenance costs are reduced, while the devices become more robust and reliable.

In this article, we have reviewed various energy harvesting techniques, and some of the ambient energy types that they use. We have seen how the core harvesting solution must be complemented by steps to minimise the node electronics’ power demand, and by providing suitable battery- or supercapacitor-based energy storage capability.