«Carles Gómez Josep Paradells José E. Caballero Edita: Fundación Vodafone España Autores: Carles Gómez Montenegro* Universitat Politècnica de ...»
Besides data processing, memory power consumption may be another contributor to energy consumption. While power for RAM is almost negligible, Flash memory writing and erasing is expensive. For example, in Crossbow MICA motes, reading and writing consume 1.1 nAh per byte and
83.3 nAh per byte, respectively.
14.1.3. Power consumption modes
A key technique in power consumption minimization in WSNs is to keep sensor nodes at low power consumption modes when no activity requires them to be at full operation. Some typical consumption modes of microcontrollers include ‘active’, ‘idle’ and ‘sleep’ modes. Regarding the radio, ‘on’ and ‘off’ states apply for the transmitter, the receiver, or both.
However, multiple consumption modes may exist, depending on the hardware platform under consideration . For instance, Texas Instruments MSP 430 defines four different sleep modes. ATMEL ATMega defines up to six modes. Different modes generally correspond to different parts of chip that can be switched off (timers, interruptions, etc.). Some modes lead to loss of part of the data stored in the RAM.
14.1.4. Communication vs. computation energy consumption
While it is difficult to perform a strict comparison between communication and computation energy consumption of a sensor node, several attempts leading to reference results are available in the literature. In particular, the energy ratio of “sending one bit” vs. “computing one instruction” has been found in the literature to be between 220 and 2900 . According to these figures, communicating (i.e. sending and receiving) one kilobyte consumes as much energy as computing three million instructions. This is a trend followed by new devices that show significant improvement in Million Instructions Per Second (MIPS) and small reduction in the transmitting power (up to an half).
In consequence, in-network processing techniques are a must in sensor networks: computation should be given as much priority as possible over communication for the sake of energy efficiency. In particular, local data processing is crucial in minimizing power consumption in multi-hop WSNs.
14.2. Power sources for sensor networks
Power sources for sensor networks can be classified into two categories:
i) energy reservoirs and ii) energy scavenging methods. Some of the most relevant technologies of each category are presented below.
14.2.1. Energy reservoirs Since it is difficult to ensure a constant power source capable of attending to the instantaneous demand from the sensor node circuit, a power storage element is recommended. This can be built based on a battery or a capacitor. Since conventional capacitors offer a very limited capacity, enhanced capacitors, known as super-capacitors or ultra-capacitors, have appeared on the market in recent years. Batteries and capacitors offer different performance and can be used individually for specific scenarios or even combined as hybrid solutions.
188.8.131.52. Primary and secondary batteries
Macro- and micro-scale electrochemical batteries have been the dominant form of power storage and delivery for electronic devices over the past 100 years. They can be considered for use as primary or secondary (i.e.
rechargeable) batteries. They have a fairly stable voltage, which precludes the extra power consumption required by power supply conditioning electronics. Table 14.4 and Table 14.5 show the energy density of three primary and secondary battery chemistries, respectively. Zinc-air batteries have the highest energy density among the chemistries considered, but they have a very short lifetime in comparison with Lithium or Alkaline batteries.
Energy density of three secondary battery chemistries Batteries are the dominant technology for powering sensor nodes.
Alkaline battery nominal shelf life is 5 years. Long life batteries can last 20 years, but they offer a voltage range from 5 to 2 Volt. Most of the sensor node chips do not support this voltage range, so the real sensor lifetime will be significantly shorter. In fact, the capacity of a battery is estimated by using a model that assumes an idealized current. In a practical situation, the sensor node voltage range is smaller than that provided by the battery, and the current required is bursty with peaks over-passing the ideal discharge current. The practical result is a shorter battery lifetime compared to the theoretical one.
All batteries suffer from a self-discharge effect. Primary batteries can lose between 8% and 20% of their capacity per year, even if they are not connected to any device. Secondary batteries lose capacity at a higher rate. For this reason, these types of batteries may limit the application lifetime without human intervention for battery replacement. The number of times a battery can be recharged depends on the technology. Low capacity Nickel Metal Hydride (NiMH) batteries can be recharged 1000 times, while high capacity ones can only be recharged 500 times.
Hydrocarbon based micro-fuel cells have very high energy densities compared to batteries. Hence, these cells could potentially be very attractive for sensor nodes requiring high power outputs for limited periods of time. One drawback, however, is the high temperatures at which energy conversion must take place and the volume of the conversion mechanism
. On the other hand, current prototypes are bulky. Nevertheless, it has been claimed that the energy density of fuel cells is ten times the energy density of comparably sized lithium-ion batteries . Another problem with micro-fuel cells is that they are based on the use of inflammable substances, which poses risks and hazards.
Ultra-capacitors store ionic charge in an electric double layer to increase their effective capacitance. Because of their increased lifetimes, short charging times, and high power densities, ultra-capacitors could be very attractive for some WSN applications . However, ultra-capacitors currently suffer from drain currents that make them unsuitable for long lifetime applications.
Fig. 14.1 shows an ultra-capacitor used in a module for WSNs.
Super-capacitors can be used when the power source is unpredictable, such as when power harvesting methods are used.
14.2.2. Energy harvesting Extracting energy from the environment and subsequently storing it for a later use is known as energy harvesting or energy scavenging. This approach is attracting more interest as low power devices are become more common. This type of power source suffers from low, variable and unpredictable levels of available power. Because of these characteristics, energy har
Chapter 14. Energy consumption and harvesting
vesting is typically used in conjunction with secondary batteries that may supply energy when scavenging energy is not possible. The sources from which energy can be harvested include ambient light, mechanical energy (movement, vibration and pressure), temperature gradients, wind flows and ambient electromagnetic (radiofrequency) radiation.
One example of a self-powered wireless sensors solution collecting and saving energy from a wide set of ambient sources is provided by EnOcean . This system makes use of energy created from slight changes in motion, light, temperature, rotation or vibration.
184.108.40.206. Photovoltaic cells
Photovoltaic technologies are commonly used to charge secondary batteries. The power density of the incident sunlight at noon on a sunny day is roughly 100 mW/cm3. Some silicon solar cells exhibit efficiencies of up to 20% . Fig. 14.2 shows an example of a node that uses photovoltaic cells.
220.127.116.11. Vibration Several researchers have developed devices to scavenge power from vibrations. Although the power signal obtained from vibrations requires significant conditioning to be useful in electronics, low level mechanical vibra
tions are present in many environments and can provide on the order of hundreds of microwatts per cubic centimetre, which is quite competitive compared to other power scavenging sources. There are three methods to produce energy with vibrational harvesters: electromagnetic (inductive) ones, electrostatic (capacitive) ones and piezoelectric ones. Current products are bulky and expensive [7, 8, 9] (see Fig. 14.3) and those based on piezoelectric effect exhibit a short life. Additionally, they should be tuned to the vibration frequency for maximum performance.
Fig. 14.3. Example of energy harvesting using vibration 18.104.22.168. Temperature gradients Energy can be scavenged from the environment when temperature variations occur. The efficiency of power conversion is proportional to the temperature variation. The power available is modest; a micro-engineered device reported below 1 µW/cm3 for a difference of temperature of 10º K . The main difficulty of the technology when applied to sensor nodes is the small size of the devices, which makes it difficult to obtain a significant thermal gradient. Nevertheless, a thermal harvesting platform that has been demonstrated to drive an IEEE 802.15.4 node is shown below ( Fig. 14.4).
Fig. 14.4. Thermal harvesting platform able to drive an IEEE 802.15.4 sensor node with a thermal difference of 3.5º K  22.214.171.124. Wind flows The potential power from moving air is proportional to v3, where v is the speed of the air. This seems to be a good option for several applications where wind can be exploited. However, wind energy harvesters are bulky and typically exhibit lower energy conversion efficiency than solar ones .
An example of a wind energy harvester is shown in Fig. 14.5.
126.96.36.199. Electromagnetic radiation This energy harvesting method is based on the use of existing ambient electromagnetic radiation available in the spectrum and generated for a different purpose than that of feeding the device. Energy densities of 0.1 µW/cm2 and 1 µW/cm2 have been reported for GSM and WiFi signal radiation, respectively . On the other hand, the most widely used power distribution method for embedded devices (besides wires) is RF radiation. Many passive devices such as RFID tags and smart tags are powered by an energy source that transmits RF energy to the passive device . Such power supply requires the use of infrastructure, and the supply source should also be supplied with power.
This approach leads to the concept of Wireless Passive Sensor Networks (WPSNs), whereby the energy for sensor nodes is supplied via RF from an external source . One advantage of this scheme when compared with other energy harvesting methods (which mainly rely on unpredictable ambient energy sources) is the fact that power supply can be controlled.
Fig. 14.6 shows a device that supplies power to sensor nodes via RF.
REFERENCES E. Shih et al., “Physical Layer Driven Protocols and Algorithm Design for Energy-Efficient Wireless Sensor Networks”, ACM Mobicom, Rome, Italy, July 2001.
 Z. Pei, Z. Deng, B. Yang, X. Chang, “Application-Oriented Wireless Sensor Network Communication Protocols and Hardware Platforms: a Survey”, IEEE International Conference on Industrial Technology, Chengdu, China, April 2008.
 “Measuring power consumption with CC2430 and Z-stack”, Application Note 053, Texas Instruments.
 Wang, A. Chandrakarasan, “Energy Efficient DSPs for Wireless Sensor Networks”, IEEE Signal Processing Magazine, July 2002.
 I.F. Akyildiz, “Wireless Sensor Networks”, 2007. Website:
http://www.ece.gatech.edu/research/labs/bwn/ee8863/ece8863.html  H. Karl et al., “Power Sources for Wireless Sensor Networks”, European Workshop on Wireless Sensor Networks (EWSN) 2004.
 KCF Technologies. Vibration Power Harvesting for Wireless Sensors.
 MIDE Piezoelectric Energy Harvesters. http://www.mide.com/products/volture/volture_catalog.php  Perpetuum (Vibration Power Harvesting). Website: http://www.perpetuum.co.uk  D. Friedman, H. Heinrich, D.-W. Duan, “A Low-Power CMOS Integrated Circuit for Field-Powered Radio Frequency Identification”, IEEE SolidState Circuits Conference, 1997.
 M. Strasser et al. “Miniaturized thermoelectric generators based on poly-Si and poly-SiGe surface micromachining”, Sensors & Actuators, 2002.
 Jennic Wireless Microcontrollers. Applications. Thermal Energy Harvesting.
 S. Sudevalayam, P. Kulkarni, “Energy Harvesting Sensor Nodes: Survey and Implications”, Technical Report, December 2008.
 O. B. Akan, M. T. Isik, B. Baykal, “Wireless Passive Sensor Networks”, IEEE Communications Magazine, August 2009.
 http://www.powercastco.com/  http://www.reghardware.co.uk/2008/10/08/toshiba_restates_– fuel_cell_goal/  EasySen website: http://www.easysen.com/products.htm  M. Raju, “ULP meets energy harvesting: A game-changing combination for design engineers”, Texas Instruments, 2008.
 http://www.enocean.com Localization techniques and wireless sensor networks Chapter 15. Localization techniques and wireless sensor networks