Ultralow-power Sensor Node with Nanowatt Wireless Circuit Technology

1. Introduction

The cabability of wireless sensor networks to gather and analyze environmental information such as vibration, temperature, angles, or positions will enable the provision of new services. Such network services, generally referred to as M2M (machine-to-machine) or the Internet of things (IoT), are attracting a great deal of attention. For example, a current sensor attached to electrical appliances in a home can tell us how much electricity is used or saved over time, and it will enable us to control appliances automatically. Sensing vibration patterns caused by wind can indicate cracks in concrete structures of bridges or buildings.

A wireless sensor network consists of wireless sensor nodes (WSNs) that can be mounted on anything and can monitor environmental or physical conditions. An image of how cargo in transport is monitored is shown in Fig. 1. WSNs attached to the cargo or to transport vehicles monitor temperature, location, or physical shocks. Then, master receivers connected to a network gather information from the WSNs and send it to servers via public networks such as Wi-Fi^*1 or 3G/4G (third generation/fourth generation). The servers store and analyze the data to produce real-time distribution schedules or to maintain quality assurance.

Fig. 1. Wireless sensor network to monitor cargo in transport.

One proposed requirement for WSNs is that they be much smaller in size and lower in cost than whatever they are attached to. Another is that they have batteries if the objects they are attached to do not. However, this last requirement is difficult to meet. Smaller batteries tend to die rather quickly, so we may need to replace them often, which would raise maintenance costs. To avoid this, WSNs should be able to operate on one battery for as long as the item they are mounted on lasts, say about 10 years, without maintenance. In some situations, it may be possible to use ambient energy sources such as solar energy, temperature, or vibration, to charge batteries.

In this article, we introduce ultralow-power circuit technology for WSNs as a way to achieve long-term WSN operation with a small battery, and we compare the power consumption of our WSNs to that of a conventional wireless system.

2. Ultralow-power wireless technology

Fabricated WSNs with ultralow-power wireless technology are shown in Fig. 2. A 5 × 5 × 5-mm WSN with a solar battery and vibration sensor is illustrated in Fig. 2(a). It transmits identifications (IDs) when the vibration sensor detects a sufficient physical shock. We confirmed in a marketing demonstration that this type of WSN was effective for monitoring which products in a store were most popular by counting the number of times products were removed from store shelves (for purchase or closer examination by customers) [1]. An evaluation kit of WSN with vibration and temperature sensors and a connector for connection to other sensors is shown in Fig. 2(b). WSNs of this type can transmit sensing data at the timing of the sensor’s trigger and timer. We demonstrated the ability of such sensors to monitor indoor temperatures and human heartbeats [2].

Fig. 2. Compact and ultralow-power WSNs.

Both types of WSNs feature the large-scale integration of nanowatt wireless circuits. The power consumption of the wireless transmitter from startup to the end of data-block transmission [3] is shown in Fig. 3. We use short impulse signals instead of a continuous sinosidal wave—the conventional method—to transmit “1” and “0” binary data. The pulses are generated by a ring oscillator, and when they stop coming, the wireless circuits also stop. This intermittent-driving technology enables us to reduce the total wireless circuit activation time and achieve ultralow power.

Fig. 3. Low-power technology.

In what follows, we explain the driving energy of WSNs and compare our technology with that of EnOcean^*2, a widely known low-power wireless product [4]. EnOcean uses a mechanical energy generator; when the button is pushed, energy is generated and IDs are sent wirelessly into the air. It is popular in Europe, where it is used in buildings to control light switches without electrical wiring.

3. Power consumption of WSNs

A block diagram of our WSN is shown in Fig. 4. The WSN comprises a wireless transmitter, digital-signal-processing (DSP) controller, sensor, and battery. Here we consider the use of a small lithium battery. The lifetime, L, of the battery is expressed as follows:

L [h] = C [mAh] ÷ I [mA] × SF,

where C is the energy capacity of the battery, I is the average current consumption, and SF is its safety factor, which corresponds to drops in battery quality due to aging or contamination. For example, if we use a coin-type battery such as a CR2032 with an average current of 1 mA, the battery lifetime is about six days; with an average current of 10 µA, the lifetime is 1.7 years.

Fig. 4. Block diagram of WSN.

Next we look at the WSN’s power consumption. The power consumption changes with the driving condtions. The driving conditions can be said to comprise the following four states:

Now let us consider the periodic transmission of sensing data. In state 1, most of the components of each block are inactive, and only the minimum necessary elements are activated, for example, the RTC (real-time clock), which counts the periods, and regulators that supply voltage. Some current leakage occurs in this state. The WSN must wake up the DSP controller in order to carry out sensing and transmitting. At this time, the WSN enters state 2 for a while to stabilize the oscillator and to read data stored in memory. Next, the DSP controller instructs the sensor to start sensing to acquire data. This requires state 2 in order to wake-up the sensor block. Then, the state will change to state 3. After data have been acquired, the system enters state 2 again to wake up wireless circuits, and the wireless transmission sequence starts. After state 4, the driving condition returns to state 1. The power consumption of each state is shown in Fig. 5.

Fig. 5. Power consumption of each state.

We plot the average power consumption, taking all states into consideration, for transmitting data versus the transmission interval for EnOcean and our ultralow-power technology [5] in Fig. 6. In the EnOcean system, 12-byte data are transmitted three times per cycle. The plot for EnOcean includes the power of the standard sensor and the voltage. When the interval is long, the static energy, corresponding to state 1, is dominant in the total power consumption, and the power becomes constant. In contrast, when the interval becomes shorter, the power consumption increases in proportion to the driving energy of each circuit. EnOcean products are specialized for a button-type power generator, i.e., triggered driving. When they are used periodically, they are useful in long intervals of over 100 seconds.

Fig. 6. Average power consumption versus transmitting interval.

The power consumption for our ultralow-power technology is plotted for the same condition as that for EnOcean—12-byte data with three transmissions per cycle. We consider two cases: using a commercially available micro-controller to transmit sensor data (case 1) and using simple DSP to transmit only IDs (case 2). The power consumption for a one-second interval is shown in Fig. 7. When the interval is short, the differences between EnOcean and our technology become large because the power consumption of our wireless circuits is 1/100 that of the EnOcean product. In EnOcean, the wireless block accounts for most of the power consumption. In contrast, with our technology, the dominant factor in the power consumption for transmitting sensing data is the power used for sensing. Thus, our technology can reduce the total power of WSNs. Moreover, if the system needs only ID information, we can take advantage of the much smaller power consumption of DSP for futher power saving. For shorter intervals, we have to further reduce the power consumption of every block.

Fig. 7. Average power consumption of one-second transmitting interval.

When the average power consumption becomes higher than 100 µA, we must use a large energy source such as a commercial power supply instead of a battery (Fig. 6). When the transmitting interval is longer than 100 seconds, the conventional approach, such as that of EnOcean, becomes useful. When the interval is 0.1 to 10 seconds, our technology has substantial advantages over the conventional technology. In such short intervals, data related to human activities, the distribution of things, and production processes are sampled continuously. This will pave the way to new applications.

References

	Shoichi Oshima Research Engineer, Ambient Device Research Group, Social Device Technology Laboratory, NTT Device Technology Laboratories. He received the B.S., M.S., and Ph.D. in electrical engineering from Tohoku University, Miyagi, in 2002, 2004, and 2008, respectively. He joined NTT Microsystem Integration Laboratories in September 2008. His research interests include wireless transceivers and circuit design for low-power communication systems. He received the Best Paper Award of IEEE Components, Packaging, and Manufacturing Technology (CPMT) Symposium Japan 2012. He is a member of the Institute of Electronics, Information, and Communication Engineers (IEICE).
	Kenichi Matsunaga Researcher, Ambient Device Research Group, Social Device Technology Laboratory, NTT Device Technology Laboratories. He received the M.E. in physical electronics from Tokyo Institute of Technology in 2010. He joined NTT Microsystem Integration Laboratories in 2010 and studies low-power complementary metal-oxide-semiconductor (CMOS) circuit design for WSNs. His research interests include RF (radio frequency) CMOS and mixed signal circuits.
	Toshihiko Kondo Senior Research Engineer, Ambient Device Research Group, Social Device Technology Laboratory, NTT Device Technology Laboratories. He received the B.E., M.E., and Ph.D. in applied chemistry from Osaka University in 1994, 1996, and 2006, respectively. He joined NTT Integrated Information and Energy Systems Laboratories in 1996 and studied hydrogen energy systems for telecommunications equipment power supplies. Since 2004, he has been engaged in research and development of communications systems and terminals for sensor network services. He has recently been studying terminals to accommodate common sensors to networks.
	Hiroki Morimura Senior Research Engineer, Supervisor, Group Leader of Ambient Device Research Group, Social Device Technology Laboratory, NTT Device Technology Laboratories. He received the B.E. in physical electronics, the M.E. in applied electronics, and the Dr.Eng. in advanced applied electronics from the Tokyo Institute of Technology in 1991, 1993, and 2004, respectively. He joined NTT in 1993. He has been engaged in the research and development of low-voltage, low-power SRAM (static random access memory) and CMOS fingerprint sensors. His research interests include noise tolerant circuit design for short-range wireless communications, ultralow-power circuit techniques for batteryless sensor nodes, and integrated CMOS-microelectromechanical systems (MEMS) technology. Dr. Morimura served on the program committee of the Symposium on VLSI Technology, 2011–2014, and on the program subcommittee on Advanced Circuits and Systems of SSDM, 2011–2013. He also served on the special section editorial committee of the IEICE Transactions on Electronics in 2011 and 2012. He received the 2004 CSS Best Paper Award, the 2006 IEICE Best Paper Award, the Best Paper Award of the Symposium on Integrated MEMS Technology at the 26th Sensor Symposium, and the Best Paper Award of IEEE CPMT Symposium Japan 2012. He is a member of IEEE (Institute of Electrical and Electronics Engineers), IEICE, and JSAP (Japan Society of Applied Physics).