Internet of Things (IoT) is one of the fastest technological trends we are seeing up to date. Millions of IoT ready devices are deployed and being connected to the network every day. We believe the total connected IoT devices will be at least 30-50 Billion up to 2020. Then, maintaining and running such an enormous cluster of IoT devices will be an extremely challenging task for us. IoT devices are expected to be running on themselves for most of their life span. Therefore, reliability is one the most important specs. Meanwhile, low power consumption is another important requirement. Relying on high capacity battery, super/ultra-capacitors or energy harvesting technology, the IoT device must be able to perform its full functionality as long as possible. From the early stage of IoT development, low-power has been one of the highest priorities in the design list. To fulfill the low power goal, the designers of IoT devices have become very careful on selecting components. Based on the useful experience of many designers and learned from lots of successful applications, we summarize following five rules of thumb that can help us achieve most of our low power IOT design goals.
In the above block diagram, the typical IoT device is partitioned into function blocks. Modern microcontrollers (MCUs) normally include the signal conditioning and ADC circuits. In some cases, we may choose to use external ADC and signal conditioning circuits due to special reasons. But typically, we use the ADC inputs come with the MCU. Therefore, an IoT node contains four parts, the MCU, sensor, wireless transceiver and power management circuit. The sensor part collects the intended environment signals, like temperature, humidity, light, magnetic field, vibration and liquid level, etc. The MCU is in charge of processing the signals sent in by the sensor and the MCU sends the processed data to the radio transceiver unit for transmission. The radio transceiver unit then transmits the data out to the relay station in the network. The power management unit provides power to each individual component on the board and ensures the voltage of the power rail is within the specs. Many IoT device are powered by non-rechargeable batteries that can last from several weeks to several years depending on the power consumption and battery capacity. In many other applications, Li-ion rechargeable batteries are used along with energy harvesting technology, which enables the system to collect energy from various sources in the ambient environment such as solar, thermoelectric, wind, RF and vibrational energy.
Utilizing energy harvesting technology, it is possible to extend the life of IoT sensor nodes to many years beyond. Since the node can obtain power from its ambient environment, which eliminates the need of replacing the battery in the sensor node, thus keep the node from maintenance. The main drawback of energy harvesting unit is that the energy sources are not steady and reliable. For example, the solar power unit may not be able to give enough output on a raining day in which case, the sensor node may have to be shut down or work at a slow duty cycle mode. Or the output of the solar unit is more than the usage demanded on a hot sunny summer day, then how to store the excessive power generated must be considered. In both above examples, energy storage will be necessary to achieve the steady supply of energy to the working components of the sensor node. The best energy storage today is for example, rechargeable Li-ion battery or supercapacitors.
For IoT device designers, it is still challenging to keep the power consumption at low level even with energy harvesting technology and high capacity batteries. If we follow the following five rules in our design of IoT devices, more efficient products can be achieved.
Rule 1: Use a low power MCU
MCU is the heart of IOT devices and it is the core in the block diagram. Being the heart of IOT devices means not only MCU is the most important component, but also it consumes the most power of the system. Today, 8-bit MCUs have become ultra-low power and the best choices for many light applications that require ultra-low power. But many IOT applications require high performance and special functions, therefore 32-bit MCUs have become the main stream choices for IOT devices thanks to the decreasing prices on some main stream 32-bit MCUS. Currently, in the market of general purpose MCUs for IOT applications, MCUs based on ARM Cortex-M series cores have become the main stream selections. The power consumption of these MCUs increases as the core number increase from M0, M0+, M3, M4 to M7. Determine the right core to use is important. The rule of thumb is that “sufficient is enough”.
There are many factors contributing to the overall power consumption of MCUs. The power consumption of the CPU core is just one of those factors. To determine the overall power consumption, we must consider at least the following three factors.
Multiple power modes that can be configurable so that the CPU core can switch to sleep mode when it does not work;
Flexible clock management so that every function block can be optimally scheduled based on need;
If there are mart low power peripheral circuits that can take part of the load of the CPU core.
In summary, the rules we are following are sleep if not working; activate if needed; avoid no-load power consumption.
In order to understand the low power characteristics of various MCUs, we have to always study the datasheets carefully to make the best choice and eventually accumulate much experience in design. Now we can also utilize the scores by ULPBench (Ultr-Low-Power Benchmark) that has been designed by EEMBC (Embedded Microprocessor Benchmark Consortium). More and people has begun to use ULPBench scores in comparing the low power characteristics of different microcontrollers, but some still keep conservative on using this tool.
Rule 2: Use integrated modular sensors
Designers usually have two opposite options in selecting sensors for their systems, discrete or integrated units. From case to case, discrete sensors may have advantages on power consumption and cost. But using integrated modular sensors may be a better choice if we take it into account of the overall power consumption of the whole system. As semiconductor technology advances, there comes the industrial trending of integrating more circuit functions into a single sensor IC design, for instance, a sensor IC that has signal conditioning and ADC circuits has become available, which sends the pre-processed signals in digital form via SPI or I2C communication ports back to the MCU, thus greatly reduces the processing load of the MCU. Some integrated sensors even have more signal processing circuits, such as a low-power CPU core, forming a sensor hub with more capable sensing function, such as completing simple calculation to reduce the load of the main MCU or simply just not to distract the MCU from processing more important tasks. All of the effects of higher integration are to further help reduce the overall power consumption of the system.
Rule 3: Select the appropriate wireless communication protocols
One of the factors that contribute to the power consumption of IoT devices is the wireless communication function. Currently there are too many wireless communication protocols in the market to select from the proprietary sub-GHz protocol to the well known Wi-Fi, BLE and ZigBee open source standards, as well as the newest developing low-power LPWAN protocols, such as LoRa, sigfox and NB-IoT, etc.. The power consumption of wireless communication may depend on the network scale, topology, reliability and data traffic rate, etc.. Generally speaking, the networks with more complicated topology and higher data rate consume more power. For example, Sub-GHz networks using star WiFi networks are more power consuming than Sub-GHz networks using peer-to-peer communication topology; similarly for ZigBee protocol, using star network saves more power than using Mesh topology. Therefore, selection of the right protocols for wireless network if highly related to the type of the IoT applications. The developers should consider the current needs and future expandability together to determine the appropriate wireless techniques for the application. Then the developers can start selecting the components with desirable low-power performance.
Rule 4: Select low-power power management devices (PMIC)
The power management devices are important part of the IoT systems. No matter you select LDO (Low Drop Out) linear regulators or DC-DC converters, even PMIC’s that provide multi-rail outputs, high efficiency is of course the primary selection criteria. Moreover, the power management devices for IoT, especially for energy harvesting applications, must first have very low standby current. For example, TI and other companies have products with standby current as low as 300 nA which reduces the standby power consumption as low as possible.
Rule 5: Plan well the power budget for security
October 2016 Internet hackers hijacked many IoT devices such as webcams and started wide-spread DDoS attacks to networks in North America. The incident reminds us to not ignore the security issues of IoT devices. To IoT device, improving the security capability means more power consumption. If the MCU does not have internal or external encryption hardware, the MCU has to consume more power to perform software encryption and decryption tasks. Therefore this part of power budget for security must be taken into consideration during design stage.
Taking all the five rules into consideration during product development, we can have a clear idea of the power consumption of the IoT devices. Based on this we can design low power IoT devices that is powerful and functional to perform to meet the design goals and stay secure from Network attacks.