IEEE TRANSACTIONS ON INSTRUMENTATION AND MEASUREMENT, VOL. 62, NO. 5, MAY 2013
1183
Energy-Aware Sensor Node Design With Its Application in Wireless Sensor Networks Ruqiang Yan, Senior Member, IEEE, Hanghang Sun, and Yuning Qian
Abstract—Energy consumption remains as a major obstacle for full deployment and exploitation of wireless sensor network (WSN) technology nowadays. This paper presents the design and implementation of an energy-aware sensor node, which can help in constructing energy-efficient WSNs. An energy-efficient strategy, which aims at minimizing energy consumption from both the sensor node level and the network level in a WSN, is proposed. To minimize the communication energy consumption of the sensor node, the distance between the transmitter and the receiver is estimated before available transmission, and then, the lowest transmission power needed to transmit the measurement data is calculated and determined. The sensor nodes are also set to sleep mode between two consecutive measurements for energy saving in normal operating conditions. Furthermore, energy saving can be achieved by estimating the energy consumption within the whole network under different network configurations and then by choosing the most energy-efficient one. Index Terms—Energy efficiency, periodic sleep/wake-up scheme, received signal strength indication (RSSI), wireless sensor network (WSN).
I. I NTRODUCTION
R
APID DEVELOPMENT of system miniaturization, wireless communication, and on-chip signal processing has promoted the development of wireless sensor technology [1], [2], which has enabled its wide applications from conditionbased maintenance [3] to industrial system monitoring [4] and environmental sensing [5]. The number of wireless sensors, which are typically considered as a wireless sensor network (WSN), deployed for real-life applications has rapidly increased in recent years, and this trend is expected to even more increase in the next years [6], [7]. However, energy consumption still remains as a major obstacle for the full deployment and exploitation of this technology, although batteries can be recharged, e.g., through solar-energy-harvesting mechanisms [8]. Prior researches have studied different approaches, such as duty-cycling and data-driven approaches [9], for reducing energy consumption. Duty cycling can be achieved through sleep/
Manuscript received June 22, 2012; revised September 3, 2012; accepted October 10, 2012. Date of publication March 7, 2013; date of current version April 3, 2013. This work was supported by the Program for New Century Excellent Talents in University under award NCET-09-0297. The Associate Editor coordinating the review process for this paper was Dr. Deniz Gurkan. The authors are with the School of Instrument Science and Engineering, Southeast University, Nanjing 210096, China (e-mail:
[email protected]). Color versions of one or more of the figures in this paper are available online at http://ieeexplore.ieee.org. Digital Object Identifier 10.1109/TIM.2013.2245181
wake-up protocols and media access control protocols with low duty cycle. For example, sparse topology and energy management approach has been proposed to improve the network lifetime by setting some redundant nodes to sleep mode [10]. The traffic-adaptive medium access protocol has been designed to reduce energy consumption by allowing sensor nodes to assume a low-power idle state whenever they are not working in transmission or receiving mode [11]. Data-driven approaches can be divided into two different categories: data compression and energy-efficient data acquisition. As an example, a variable data length coding method using Walsh function was developed to compress the transmission data, and this has been proved to be effective in improving energy efficiency in signal transmission [12]. In another study, the sensor network was divided into several subsystems, and only high-level inferences are communicated between the subsystems. In this way, the energy consumption for communication decreases as the data to be transmitted decrease [13]. For energy-efficient data acquisition, an adaptive sampling algorithm consisting of duty cycling (the sensor board is switched off between two consecutive samples) and adaptive sampling (the optimal sampling frequency is estimated online) is proposed to reduce energy consumption in a sensor network [14]. Researchers have also studied other approaches for energy-aware transmission, including modulation scaling schemes [15], [16], multihop routing schemes [17], network sectioning [18], [19], and low-power hardware [20]. Furthermore, a combination of sleep scheduling with block transmission approach has been proposed to achieve energy saving in a wireless multimedia sensor network [21]. Motivated by the prior research, an energy-saving strategy consisting of node-level energy saving using adaptive radio frequency (RF) power setting and network-level energy saving through adaptive network configuration has been proposed in our conference paper [22]. This paper is an extension of [22], in which the periodic sleep/wake-up scheme is added into the sensor node design to further achieve the node-level energy saving. The whole communication procedure and experimental test has also been updated with the new ultralow-power microcontroller MSP430F149 being used as the core for the designed sensor node. The remainder of this paper is organized as follows. Section II introduces the sensing scheme and the associated energy consumption estimation of the sensor node and the entire network, respectively. Section III presents the hardware design of the sensor node, followed by energy-saving scheme implementation. In Section IV, the performance of the presented strategy is evaluated in an experimental sensor network. The conclusion is finally drawn in Section V.
0018-9456/$31.00 © 2013 IEEE
1184
IEEE TRANSACTIONS ON INSTRUMENTATION AND MEASUREMENT, VOL. 62, NO. 5, MAY 2013
II. S ENSING S CHEMES The concept of wireless sensor node implies that, except for the physical sensing capabilities, the nodes will also be able to process the obtained data and communicate the results wirelessly. In recent years, many energy conservation schemes have been proposed in the literature (a detailed survey can be found in [9]), which assume that data acquisition and processing have an energy consumption that is significantly lower than communication [14]. In addition, since each of the sensor nodes in the network is energy constrained and each component in a sensor node consumes a certain amount of energy, power supply becomes important to ensure proper operation of the entire WSN as the number of sensors deployed in a network grows. Hence, constructing effective network structures for the application of WSN with consideration of energy efficiency is of critical importance.
Fig. 1.
Scheme 1 used in the greenhouse.
this case is calculated as A. Energy Consumption Calculation After sensing the environmental parameters, the results should be transmitted to the central monitoring unit (CMU) or other sensor nodes. In order for two sensor nodes to communicate, the energy consumption needed for data transmission can be expressed as [23] ETx = Ee_tx · k + εamp · dα
(1)
where k is the number of transmitted data bits; α is a factor valued from 2 to 5, depending on the environment of wireless transmission; d is the distance between two sensor nodes; εamp (J/b/m2 ) is the amplification coefficient to satisfy a minimum bit error rate to ensure reliable reception at the receiver; and Ee_tx (J/b) is the energy dissipated to operate the transceiver, which is given as Ew_tx = Vcc · IT P /Kdata_rate
(2)
where Vcc denotes the working voltage, IT P denotes the current for transmission, and Kdata_rate denotes the data transmission rate. The energy consumed for receiving a data stream can be expressed as ERx = Ee_rx · k.
Edr =
N
[(Ee_tx + εamp · dα n ) · kr ]
(4)
n=1
where N is the number of sensors, dn is the distance between each sensor node and the CMU, and kr is number of data bits from the obtained data. For example, Fig. 1 shows scheme 1 being applied to the platform of greenhouse management in which temperature is measured by the sensors. In each greenhouse, the sensor nodes acquire the temperature data and transmit the data to the CMU directly without routing and relay. Scheme 2: The sensors are grouped into different clusters, and the obtained data from each sensor node are transmitted to the corresponding cluster head (it is defined as the sensor node that collects the data from others in the cluster). Then, the cluster head will pack the data and transmit them to the CMU. The energy consumption Eds in this case is calculated as
Eds =
M
N −1 m
m=1
j
Ee_tx + εamp · dα j + Ee_rx · kr + (Ee_tx + εamp · dα m ) · km
(5)
(3)
Equation (1) shows that, for a fixed distance, the energy consumed is proportional to the number of data bits. On the other hand, the longer the distance between two sensor nodes is, the more energy will be consumed. B. Sensing Schemes The network-level energy saving was realized mainly through the scheme switching of the network. Two different schemes of the network are shown as follows [13]. Scheme 1: The obtained data points are transmitted to the CMU from each sensor node. The energy consumption Edr in
where M is the number of clusters, Nm is the number of sensors in the corresponding cluster, dj is the distance between a sensor and its corresponding cluster head, dm is the distance between the cluster head and the CMU, and km is the number of packed data bits for data transmission. For example, Fig. 2 shows scheme 2 being applied to the platform of greenhouse management in which the network is grouped into different clusters. In this scheme, each greenhouse is defined as a cluster, and the obtained data are transmitted first to corresponding cluster head; then, the CMU will ask the cluster heads for the temperature data.
YAN et al.: ENERGY-AWARE SENSOR NODE DESIGN WITH ITS APPLICATION IN WIRELESS SENSOR NETWORKS
1185
scheme 2 is applied for local data transmission from a sensor node to its cluster head, the cluster head sends a test code with maximum transmission power PTx max to one of the sensor nodes in the local cluster first. By measuring the received power PRx on each sensor node, the distance to the cluster head can be calculated as dˆα ≥ PTx max /PRx .
(8)
Hence, by minimizing the estimated distance for data transmission, the minimum required power to ensure data communication is expressed as PTx = PS · Fig. 2.
Scheme 2 used in the greenhouse.
III. D ESIGN OF THE E NERGY-AWARE S ENSOR N ODE In order to provide energy-efficient sensing in a WSN, an energy-aware sensor node is designed and implemented in this section. The detailed information is described in the following discussion.
PTx max . PRx
In this paper, Chipcon CC1101 is chosen as the RF transceiver used in the sensor node design. CC1101 is a lowcost true single chip UHF transceiver designed for very low power wireless applications. The received signal power PRx in CC1101 is translated to a decibel (dBm) value—RSSI. The relationship between PRx and RSSI is defined as follows [24]: RSSI = 10 · log10 PRx .
A. Communication Module In the two sensing schemes designed for the WSN, it is assumed that the transmission power is minimized to ensure reliable reception at the receiver end, according to the communication distance between two sensor nodes. Hence, awareness of the communication power as well as the adjustability of the transmitter’s output power becomes critical in performing the sensing scheme for the designed sensor node. By assuming a unit signal gain provided by antennas, the output power of the communication module is dominated by the consumption for power amplifier. To transmit 1 bit to the receiver, the output power and associated received power are expressed as PTx = (εamp · R) · dˆα PRx =
PTx = (εamp · R) · dα
dˆα dα
= PS ·
dˆα dα
(6) (7)
where R denotes the data transmission rate, dˆ and d are the estimated and actual transmission distances between the transmitter and the receiver, respectively, and PS = εamp · R is the receiver sensitivity denoting the minimum signal power that the receiver can discern. From (7), it is seen that, if the estimated distance dˆ < d, then the received signal cannot be identified and the communication between sensor nodes fails. On the other hand, if dˆ > d (overestimation), which means a received power that is higher than receiver sensitivity, then a portion of the transmission energy will be lost on the propagation path while not affecting the results of signal reception. In this case, the energy efficiency problem is translated to the effective estimation of communication distance between two sensor nodes. Since all of the sensor nodes are equipped with both transmission and receiving capabilities, we can estimate the distance between sensor nodes through received signal strength indication (RSSI). For example, considering the case where sensing
(9)
(10)
In Rx mode, the RSSI value can be read continuously from the RSSI status register, which is a binary complement number. The following procedure can be used to convert the RSSI reading to an absolute power level (RSSIdBm )[24]. 1) Read the RSSI status register. 2) Convert the reading from a hexadecimal number to a decimal number (RSSIdec ). 3) If RSSIdec ≥ 128, then RSSIdBm = (RSSIdec − 256)/ 2 − RSSIoffset . 4) Else if RSSIdec < 128, then RSSIdBm = RSSIdec /2 − RSSIoffset where RSSIoffset is a typical value corresponding to the data rate and the center frequency. CC1101 is programmed for operation at 100 kb/s with 433 MHz; therefore, RSSIoffset = 75. From (9), we can further obtain the following relationship as 10 · log10 PTx = 10 · log10 PS + 10 · log10 PTx max − 10 · log10 PRx
(11)
where 10 · log10 PTx is the transmission power in dBm and 10 · log10 PRx is the RSSI. There is a register PATABLE in CC1101 which is used in selecting the transmission power. From (11), we can compute the minimum required power as PATABLE = 10 · log10 PS +10 · log10 PTx max −RSSI. (12) In CC1101, the format of the data packet can be configured to include the following items: 1) preamble; 2) synchronization word; 3) length byte; 4) address byte; 5) payload; 6) CRC word.
1186
IEEE TRANSACTIONS ON INSTRUMENTATION AND MEASUREMENT, VOL. 62, NO. 5, MAY 2013
Fig. 3. Data packet in this paper.
Each item can be configured by setting the corresponding registers when initializing CC1101. In this paper, the data packet is configured as shown in Fig. 3. As a result, each sensor node will transmit (10 + n) byte data one time, and the energy consumption can be computed as follows: ETx = PTx ·
(10 + n) · 8 R
(13)
where n denotes the number of payload data bits needed to be transmitted. Here, temperature data are the payload data, and R = 100 kb/s is used in this paper. B. Periodic Sleep/Wake-Up Scheme Except for the adaptive RF power supply as described in the previous section, the node-level energy saving is also achieved through the periodic sleep/wake-up scheme. As we know if a WSN is deployed in remote fields or under harsh environments [25] where manually recharging batteries for sensors is not feasible, one typical alternative approach for energy saving is to turn off some sensors and to activate only a necessary set of sensors while providing a good sensing coverage and network connectivity simultaneously [26]. In slowly varying parameter measurements, such as temperature, not all sensor nodes are needed to stay in active mode. Therefore, in order to save energy consumption, the sensor nodes are designed to be put into a sleep mode with a timer that determines their sleep duration. When the timer overflows, an interrupt happens, and it will wake those nodes up and will then perform measurements and data transmission. C. Sensor Node Design Each sensor node should be able to collect environmental parameters and communicate with each other. At the same time, its corresponding hardware and software should be energy efficient and include the functionality described in previous sections. Based on these requirements, the sensor node has been designed and implemented as shown in Fig. 4. In this design, the MSP430F149 microcontroller is chosen as the core for fulfilling computation and control functions of the sensor node; this is because it possesses the capability of ultralowpower consumption and short waking-up time (less than 6 μs) as compared to other commonly used microcontrollers listed in Table I. Also, from Tables I and II, the architecture of MSP430F149, combined with its five low-power modes (LPM0–LPM4), can be optimized to achieve extended battery life in portable measurement applications. Furthermore, when the CPU works at the same speed (1 MHz), the energy con-
Fig. 4.
Sensor node developed for temperature monitoring.
sumption of MSP430F149 is lower than that of other commonly used microcontrollers (shown in Table I). The supply voltage of the sensor node is 3.3 V, which is obtained using the AMS1117-3.3 voltage regulator with low intrinsic consumption current from two 3.7-V (2400-mAh) lithium batteries. The RS232 interface is designed for communication with the PC in environment monitoring. For example, when collecting the temperature parameters, these temperature values can be transmitted to PC and displayed in the computer monitor. D. Network-Level Energy-Saving Realization Although the energy consumption of each sensor node is individually minimized by its associate functional modules, the total energy consumption can be further reduced by using the appropriate sensing scheme for the whole sensor network. First, the sensor node is set to sleep mode until the timer overflows, and then, it is waked up to collect environmental parameters and waits to communicate with the CMU. Second, the proposed scheme initially estimates the minimum required transmission power PTx by using the test code from the CMU or the cluster head. Then, the CMU selects an appropriate sensing scheme by comparing the total energy consumption. By assuming a homogeneous hardware scheme for all of the sensor nodes, as shown in Fig. 4, the variables representing distances between sensor nodes/cluster heads and CMU are aggregated by broadcasting request to all of the sensor nodes. After all information is collected, the CMU estimates overall energy consumption for each of the schemes and makes decision to choose the one with the best energy efficiency as the current network scheme. In this scheme, test code is defined as the command for sensor nodes to implement different tasks. Here, test code 1 informs all of the sensor nodes to estimate patable1, test code 2 informs all of the cluster heads to broadcast test code 3 in the local cluster, test code 3 informs the sensor nodes in the local cluster to estimate patable2, test code 4 informs all of the nodes to select patable1 as its transmission power and to set the CMU ID as its destination address, and test code 5 informs all of the nodes to select patable2 as its transmission power and to set its cluster head ID as its destination address.
YAN et al.: ENERGY-AWARE SENSOR NODE DESIGN WITH ITS APPLICATION IN WIRELESS SENSOR NETWORKS
1187
TABLE I C OMPARISON A MONG D IFFERENT M ICROCONTROLLERS
TABLE II C URRENT C ONSUMPTION OF E ACH M ODE FOR MSP430F149
Fig. 6. Flowchart of the main program for timer initialization and low-power mode setting.
3. 4. 5. 6. 7. 8. 9. 10. 11. 12. 13. 14. 15. 16. 17. 18. 19. Fig. 5.
Flowchart for energy estimation and scheme selection on the CMU.
Fig. 5 shows the flowcharts for energy-related information gathering, potential energy consumption calculation, and sensing scheme switching. The corresponding proposed algorithm is given in Algorithm 1. Algorithm 1: CMU_Estimate(M, N ) 1. PATABLE| = 0 × xc0; //maximum transmission power 2. For (i = 0; i < M ; i++){
Load the address of destination sensor node node_ID[i]; Send TestCode1; While(ReceivedF lag == 0); ETx(S1)[i] = 1010·log10 PS +10·log10 PTx max −RSSI[i] /10 ; } ETx(Sl) = M i=1 ETx(Sl[i]) ; For(j = 0; j < N ; j++){ Load the address of destination cluster head ClusterHead_ID[j]; Send TestCode2; While (ReceivedF lag == 0); Hj 10·log PS +10·log PTx max −RSSI/10 10 10 10 · ETx(S2)[j] = i=0 ((10 + n) · 8/R); } ETx(S2) = N j=1 ETx(S2)[j] ; If (ET x(S1) < ET x(S2)) Broadcast TestCode4; Else Broadcast TestCode5;
In Algorithm 1, M and N denote the number of sensor nodes and clusters in the network, respectively. The array N ode_ID[i] and ClusterHead_ID[j] are defined as buffers to save the address of the sensor nodes and cluster heads, respectively. ReceivedF lag is defined as a flag to indicate the receiving state (ReceivedF lag = 0 indicates that the sensor node does not receive any data bits, and ReceivedF lag = 1 indicates that the sensor node receives all data.) Fig. 6 shows the main program of the sensor node and the cluster head for timer initialization and sleep mode setting.
1188
IEEE TRANSACTIONS ON INSTRUMENTATION AND MEASUREMENT, VOL. 62, NO. 5, MAY 2013
Fig. 8.
Network structure of sensing scheme 1.
Save the RSSI value of the ith cluster head in the RSSI array RSSI[H]; 16. } 17. Send the RSSI array RSSI[H] to CMU; 18. } 19. Else if(ClusterF lag == 2){ 20. While(ReceivedF lag == 0); 21. Send RSSI to the corresponding cluster head; 22. PATABLE2 = 10 · log10 PS + 10 · log10 PTx max − RSSI; 23. } 24. } 15.
Fig. 7. Flowchart for energy estimation and scheme selection steps on sensors and cluster heads.
Fig. 7 shows the flowchart of the interrupt service routine of the sensor node and the cluster head for estimating the minimum required transmission power and the scheme switching. The corresponding proposed algorithm is given in Algorithm 2. Algorithm 2: Sensor_Estimate_ISR() 1. While(ReceivedF lag == 0); 2. If(Rx_TestCode==TestCode1){ 3. Send RSSI to CMU; 4. PATABLE1 = 10 · log10 PS + 10 · log10 PTx max − RSSI; 5. } 6. Else if(Rx_TestCode==TestCode2){ 7. If(ClusterF lag == 1){ 8. Save the RSSI value of cluster head in the RSSI array RSSI[H]; 9. For(i = 1; i < H; i++){ 10. Save the RSSI value of cluster head in the RSSI array RSSI[H]; 11. PATABLE| = 0 × c0; //maximum transmission power 12. Load the address of destination cluster node ClusterN odeI D[i]; 13. Send TestCode3; 14. While(ReceivedF lag == 0);
In Algorithm 2, ClusterF lag is a status flag used to indicate whether the sensor node is cluster head. ClusterN ode_ID[i] is defined as a buffer to save address of the sensor nodes in clusters. RSSI[H] is used as a buffer to save the RSSI values of the sensor nodes in the clusters. IV. E XPERIMENTAL T EST In this paper, the energy-efficient strategy that we designed on the sensor platform consists of two parts: node-level energy saving and network-level energy saving. The node-level energy saving is achieved by adaptive transmission power setting and by the periodic sleep/wake-up scheme. The network-level energy saving is achieved by adaptive network configuration. In the experimental test, we first investigate the performance of node-level energy saving. Then, the performance of networklevel energy saving is investigated by comparing the energy consumption of fixed network configuration with that of adaptive network configuration. A. Traditional Versus Node-Level Energy-Saving Scheme In this experimental test, we compared the communication energy consumption of fixed transmission power setting (which we call as traditional scheme) with that of adaptive transmission power setting when sensing scheme 1 is used in the network, as shown in Fig. 8. Then, the performance of the periodic sleep/wake-up scheme was further investigated by comparing the energy consumption of one sensor node among several working modes.
YAN et al.: ENERGY-AWARE SENSOR NODE DESIGN WITH ITS APPLICATION IN WIRELESS SENSOR NETWORKS
1189
TABLE III O UTPUT P OWER AND C URRENT OF CC1101
Fig. 10. Communication energy consumption of the sensor network with different scales.
Fig. 9. Communication energy consumption under different transmission power settings.
The proposed node-level energy saving consists of adaptive transmission power setting and periodic sleep/wake-up scheme. The relationship between output power and current of the transceiver CC1101 is shown in Table III [24]. It is known that the transmission power was maximized to ensure reliable communication on traditional WSN. Here, the output power of CC1101 is always set to 10 dBm. The adaptive transmission power scheme allows sensor nodes to adjust their output power based on the estimated distance between the sensor node and the CMU. Since wireless communication is always affected by several environmental factors, such as temperature and humidity, we repeated the experimental test for 50 times to ensure the accuracy and reliability. Fig. 9 illustrates the communication energy consumption of one sensor node under the two settings with 1-m distance communication. The blue curve illustrates the communication energy consumption of fixed transmission power setting, and the red one illustrates the energy consumption of adaptive transmission power setting. It can be seen that, for a fixed distance, the energy saving achieves up to 50% when adaptive transmission power setting is applied. Fig. 9 only shows communication energy consumption for one sensor node, while Fig. 10 illustrates communication energy consumption of the whole network. It can be seen that the energy saving increases as the number of sensor nodes increases. In summary, the node-level energy-saving scheme is more energy efficient than the traditional one, and as the
Fig. 11. Energy consumption of one sensor node under different schemes
Fig. 12. Network structure of sensing scheme 2.
number of sensor nodes increases, this effect will be more obvious. In order to investigate the performance of the periodic sleep/wake-up scheme, we further compared the energy consumption of the sensor node by including the energy consumption of the CPU itself. The proposed periodic sleep/wake-up
1190
IEEE TRANSACTIONS ON INSTRUMENTATION AND MEASUREMENT, VOL. 62, NO. 5, MAY 2013
Fig. 14.
Comparison between schemes 1 and 2 under different distances.
TABLE IV S TATISTICAL R ESULTS OF C OMMUNICATION E NERGY C ONSUMPTION
to stay in the active mode for about 0.5 s to perform default measurement and communication. Then, we can calculate the energy consumption of MSP430F149 in one work cycle as ¯ = PL · tL + Pa · ta = (IL · tL + Ia · ta ) · U E
(14)
where IL is the current consumption of LPM0 (55 μA), tL is the sleep time of MSP430F149 (2 s), Ia is the current consumption of the active mode (160 μA), and ta is the time of one work cycle (0.5 s). Fig. 11 illustrates the total energy consumption of one sensor node under four different working modes. It can be seen that the proposed periodic sleep/wake-up scheme achieves more energy savings. B. Scheme 1 Versus Scheme 2
Fig. 13. Communication energy consumption of scheme 1 and scheme 2. (a) D = 1 m. (b) D = 5 m. (c) D = 10 m.
scheme allows sensor nodes to be set to sleep mode during the measurements and to be waked up by an interrupt event. Here, on the platform of MSP430F149, a timer is used to set the time interval at 2 s. At the same time, MSP430F149 is set to LPM0 mode until the timer overflows and wakes the sensor node up
In this experimental test, the sensor nodes are deployed as shown in Fig. 12 to evaluate the performance of the adaptive network configuration scheme. Communication energy consumption was estimated with different distances. The results are shown in Fig. 13. In traditional WSNs, the network configuration will never change after being initialized, even though the deployment of the sensors is changed. The blue curve in Fig. 13 illustrates the communication energy consumption of the whole network when the sensor nodes are deployed as shown in Fig. 12 and when scheme 1 is applied. In contrast, the red curve illustrates the communication energy consumption of the whole network when scheme 2 is applied. It can be concluded that scheme 2 is more suited for such a network as shown in Fig. 12 than scheme 1, as it is more energy efficient. Fig. 14 and Table IV provided the mean value of the communication energy consumption. It can be seen that the adaptive network configuration scheme achieves more and more energy savings as the communication distance between the CMU and the cluster heads increases.
YAN et al.: ENERGY-AWARE SENSOR NODE DESIGN WITH ITS APPLICATION IN WIRELESS SENSOR NETWORKS
V. C ONCLUSION In this paper, we have presented the design and implementation of an energy-aware sensor node, which can help in constructing an energy-efficient WSN through “node-level energy saving” and “network-level energy saving.” The “nodelevel energy saving” is achieved by adaptive transmission power setting and by the periodic sleep/wake-up scheme, while the “network-level energy saving” is achieved by adaptive network configuration. The experimental tests have confirmed the effectiveness of the presented schemes for energy saving in a WSN. For further study, more functionality, such as local data processing and data fusion algorithm, will be added into the designed sensor node for improving the energy-saving protocol. In addition, the real-world applications, e.g., greenhouse management, of such a WSN will be explored. R EFERENCES [1] C. Chong and S. P. Kumar, “Sensor networks: Evolution, opportunities, challenges,” Proc. IEEE, vol. 91, no. 8, pp. 1247–1256, Aug. 2003. [2] R. Gao and Z. Fan, “Architectural design of a sensory node controller for optimized energy utilization in sensor networks,” IEEE Trans. Instrum. Meas., vol. 55, no. 2, pp. 415–428, Apr. 2006. [3] A. Tiwari, P. Ballal, and F. L. Levis, “Energy-efficient wireless sensor network design and implementation for condition-based maintenance,” ACM Trans. Sens. Netw., vol. 3, no. 1, pp. 1–23, Mar. 2007. [4] F. Salvadori, M. Campos, P. Sausen, R. Camargo, C. Gehrke, C. Rech, M. Spohn, and A. Oliveira, “Monitoring in industrial systems using wireless sensor network with dynamic power management,” IEEE Trans. Instrum. Meas., vol. 58, no. 9, pp. 3104–3111, Sep. 2009. [5] D. Gallo, C. Landi, and N. Pasquino, “Multisensor network for urban electromagnetic field monitoring,” IEEE Trans. Instrum. Meas., vol. 58, no. 9, pp. 3315–3322, Sep. 2009. [6] ON World Inc., Wireless Sensor Networks: Growing Markets, Accelerating Demands, Jul. 2005. [Online]. Available: http://www.onworld.com/ html/wirelesssensorsrprt2.htm [7] Embedded WiSeNTs Consortium, Embedded WiSeNTs Research Roadmap (Deliverable 3.3). [Online]. Available: http://www. embedded-wisents.org/ [8] C. Alippi and C. Galperti, “An adaptive system for optimal solar energy harvesting in wireless sensor network nodes,” IEEE Trans. Circuits Syst. I, Reg. Papers, vol. 55, no. 6, pp. 1742–1750, Jul. 2008. [9] A. Giuseppe, C. Marco, and D. F. Mario, “Energy conservation in wireless sensor networks: A survey,” Ad Hoc Netw., vol. 7, no. 3, pp. 537–568, May 2009. [10] C. Schurgers, V. Tsiatsis, and M. B. Srivastava, “STEM: Topology management for energy efficient sensor networks,” in Proc. IEEE Aerosp. Conf., 2002, pp. 3-1099–3-1108. [11] V. Rajendran, K. Obraczka, and J. J. Garcia-Luna-Aceves, “Energyefficient, collision-free medium access control for wireless sensor networks,” Wireless Netw., vol. 12, no. 1, pp. 63–78, Feb. 2006. [12] A. Kadrolkar, R. Gao, and R. Yan, “Energy efficient data transmission for manufacturing system health monitoring,” in Proc. 9th Int. Conf. Frontiers Des. Manuf., Changsha, China, Jul. 2010, pp. 65–70. [13] R. Yan, D. Ball, A. Deshmukh, and R. Gao, “A Bayesian network approach to energy-aware distributed sensing,” in Proc. IEEE Sens. Conf., Vienna, Austria, Oct. 2004, pp. 44–47. [14] C. Alippi, G. Anastasi, M. D. Francesco, and M. Roveri, “An adaptive sampling algorithm for effective energy management in wireless sensor networks with energy-hungry sensors,” IEEE Trans. Instrum. Meas., vol. 59, no. 2, pp. 335–344, Feb. 2010. [15] V. Raghunathan, C. Schurgers, and S. Park, “Energy-aware wireless microsensor networks,” IEEE Signal Process. Mag., vol. 19, no. 2, pp. 40– 50, Mar. 2002. [16] C. Schurgers, V. Raghunathan, and M. Srivastava, “Power management for energy-aware communication systems,” ACM Trans. Embedded Comput. Syst., vol. 2, no. 3, pp. 431–447, Aug. 2003. [17] E. Biyikoglu, B. Prabhakar, and A. Gamal, “Energy efficient packet transmission over a wireless link,” IEEE/ACM Trans. Netw., vol. 10, no. 4, pp. 487–499, Aug. 2002.
1191
[18] R. Gao, A. Deshmukh, R. Yan, and Z. Fan, “Energy efficient wireless sensor network for dynamic system monitoring,” in Proc. SPIE, 2005, vol. 5999, no. 4, pp. 1–10. [19] D. Ball, R. Yan, T. Licht, A. Deshmukh, and R. Gao, “A strategy for decomposing large-scale energy-constrained sensor networks for system monitoring,” Product. Plann. Control, vol. 19, no. 4, pp. 435–447, 2008. [20] A. Wang, W. Heinzelman, A. Sinha, and A. Chandrakasan, “Energyscalable protocols for battery-operated microsensor networks,” J. VLSI Signal Process. Syst. Signal Image Video Technol., vol. 29, no. 3, pp. 223– 237, Nov. 2001. [21] C. Sha, R. Wang, H. Huang, and L. Sun, “An energy-saving strategy based on sleep scheduling and block transmission for wireless multimedia sensor networks,” Int. J. Pervas. Comput. Commun., vol. 6, no. 2, pp. 248– 267, 2010. [22] H. Sun, Y. Qian, and R. Yan, “Design and realization of an intelligent sensor node with its application in energy-aware WSNs,” in Proc. IEEE I2MTC, Graz, Austria, May 11–13, 2012, pp. 941–946. [23] A. Wang and A. Chandrakasan, “Energy-efficient DSPs for wireless sensor networks,” IEEE Signal Process. Mag., vol. 19, no. 4, pp. 68–78, Jul. 2002. [24] Texas Instruments, CC1101 Data Sheet 2010. [Online]. Available: www. ti.com.cn [25] A. Chehri, P. Fortier, and M. Tardif, “UWB-based sensor networks for localization in mining environments,” Ad Hoc Netw., vol. 7, no. 5, pp. 987–1000, Jul. 2009. [26] G. Nan, G. Shi, Z. Mao, and M. Li, “CDSWS: Coverage-guaranteed distributed sleep/wake scheduling for wireless sensor networks,” EURASIP J. Wireless Commun. Netw., vol. 44, pp. 762–776, Feb. 2012.
Ruqiang Yan (M’07–SM’11) received the M.E. degree from the University of Science and Technology of China, Hefei, China, in 2002 and the Ph.D. degree from the University of Massachusetts Amherst, Amherst, MA, USA, in 2007. He joined the School of Instrument Science and Engineering, Southeast University, Nanjing, China, as a Full Professor in October 2009. His research interests include nonlinear time-series analysis, multidomain signal processing, and energy-efficient sensing and sensor networks for condition monitoring and health diagnosis of large-scale, complex, and dynamical systems. Dr. Yan is a member of the ASME. He is an Associate Editor of the IEEE T RANSACTIONS ON I NSTRUMENTATION AND M EASUREMENT. He is a Cochair of the Technical Committee on Signals and Systems in Measurement of the IEEE Instrumentation and Measurement Society. He received the New Century Excellent Talents in University Award from the Ministry of Education of China in 2009.
Hanghang Sun received the B.E. degree from Zhengzhou University, Zhengzhou, China. He is currently working toward the M.Eng. degree in instrument science and technology at Southeast University, Nanjing, China. His research interests include energy-efficient wireless sensor networks for condition monitoring of mechanical systems and environmental monitoring.
Yuning Qian received the B.E. degree from Southeast University, Nanjing, China, where he is currently working toward the Ph.D. degree in the School of Instrument Science and Engineering. His research interests include wireless sensor networks, nonlinear time-series analysis, and mathematical modeling for health diagnosis and remaining-life prediction of mechanical systems.
