ENERGY MANAGEMENT OF WIRELESS SENSOR NETWORK WITH MIMO TECHNIQUES

Transcription

1 EERGY MAAGEMET OF WIRELESS SESOR ETWORK WITH MIMO TECHIQUES TRUPTI MAYEE BEHERA Schoo of Eectronics, KIIT University Bhubaneswar, Odisha,India SUDHASU SEKHAR SIGH Schoo of Eectronics, KIIT University, Bhubaneswar, Odisha,India Abstract : With recent advances in depoyment of sensor nodes mounted on mobie patforms, node mobiity is becoming an attractive aternative to improve network coverage dynamicay in sensor networks. Mobie sensor networks depoyed for surveiance appications, it is important to use an efficient energy management scheme that can empower nodes to make better decisions regarding their positions such that strategic tasks ike target tracking can be benefited from node movement. However, due to energy constrain nodes; it may not be cost effective to depoy a arge number of mobie nodes for continuous movements. Due to their effectiveness for enhancing energy and bandwidth efficiency, MIMO schemes have been studied intensivey in recent years and used for energy conservation. Keywords Wireess sensor networks, Energy efficiency, Virtua MIMO, STBC, SCHCT 1. Introduction Energy consumption is the core issue in wireess sensor networks (WS). To generate a node energy mode that can accuratey revea the energy consumption of sensor nodes is an extremey important part of protoco deveopment, system design and performance evauation in WSs[1]. Due to the physica size and energy imitation of sma sensor nodes, directy depoyment of MIMO in one sensor node is infeasibe in practice. However, through sensor nodes cooperation, virtua MIMO technique can be impemented in sensor networks. In virtua MIMO network, a group of sensors cooperate to transmit and receive data. Space time bock coding [][3] is a technique used in wireess communications to transmit mutipe copies of a data stream across a number of antennas and to expoit the various received versions of the data to improve the reiabiity of datatransfer. Hence in this paper the STBC based custer heads cooperative transmission (SCHCT) scheme is proposed.. System Mode Consider a wireess sensor network with sensing nodes distributed randomy in a square area of side M meters. A sensor nodes are assumed to be stationary, heterogeneous and energy constrained, where each node can transmit data to any other node and sink as shown in Figure 1. The sink node is assumed to have no energy constraints and is equipped with one or more receiving antennas. The sensor nodes are geographicay grouped into receiving nodes and non-custer head nodes that sense the data from the sensing fied. The custer heads are re-eected after each round of data transmission. ISS : Vo. 4 o.06 June

2 Figure 1 The sink is assumed to be equipped with more than one antenna in order to impement the MIMO transmission. Each node has a packet of L bits to the custer head. The operation of SCHCT is divided into rounds [4]. Each round consists of three stages: the custer formation stage, the steady state stage and the cooperative transmission stage.1. Custer Formation Stage Step 1: In this stage, a the sensor nodes sef-organized to form custers. Each node in the network uses the cassic carrier sense mutipe access with coision avoidance (CSMA/CA) scheme to contend for the wireess channe [5]. Once a node x succeeds in accessing the channe, it sends a message at fixed transmission energy to discover its one-hop neighbours. This message carries the foowing information: node ID, its remaining energy, and a ist of x's neighbours (nodes that x has received messages from). In genera, a node sends messages as soon as it joins the network and whenever it hears from new neighbours. Step : The next step is to seect the Master custer head (MCH). Since MCHs do more work than any typica node (for coecting, aggregating, and forwarding data), the seection criterion of MCHs is the node's remaining energy. Each node maintains tabes of remaining energy vaues of a its 1-hop neighbours.a the nodes start the custering process.every node compares its remaining energy to those of its one-hop neighbours. A node waits for other undecided neighbours with higher remaining energy to decide before itsef. If the node has the highest remaining energy in its neighbourhood, it decares itsef as an MCH and announces that to its neighbours. Step 3: The next step is to associate a Save custer head (SCH) with each MCH, if possibe. To seect SCHs, each MCH sends an SCH invitation message to the neighbouring node whose neighbour ist overaps the most with that MCH's neighbour ist. The invited node announces its decision via an SCH acceptance message, enabing decided MCHs to ook for other SCHs. Upon receiving an SCH acceptance message, the intended MCH confirms this association via an SCH confirmation message. Step 4: After the eection of the custer heads, each custer head wi transmit a broadcast message to other sensor nodes in the area using CSMA protoco. The message contains the custer head s ID. Sensor nodes then choose one of the custer head to join in based on the signa strength of the broadcast message. By sending the join-request message to the nearest custer head, the custer head wi record the ID of the custer members. After the formation of the custer, the custer head wi set up a TDMA schedue for the custer members and then broadcast this schedue to each of the custer members.. Steady State Stage In this phase, custer members wi transmit their data to the custer head by mutipe frames as in the origina LEACH scheme. In each frame, each custer member wi transmit its data during its aocated transmission sot specified by the TDMA schedue in custer formation phase, and it wi be seep in other sots to save energy. After a custer head receives data frames from its custer members, it wi perform data aggregation to remove the redundancy in the data..3 Cooperative Transmission Stage After the aggregation, custer heads then wait for the broadcast poing message from the sink. When the custer head receives the broadcast message, it repies with an answer message. This answer message contains the ID and ocation information of each custer head. Based on this information, the sink wi group custer heads into virtua MIMO ce and set up a TDMA schedue for the virtua MIMO ce. This group and schedue information wi be broadcasted to every custer head. Then the custer heads in virtua MIMO ce wi cooperate to form the virtua antenna array. Each custer head in virtua MIMO ce first broadcasts its information to a the other custer heads in the virtua MIMO ce using different time sots [6][7]. After a custer heads receives the information bits from other custer heads in virtua MIMO ce, they encode the transmission sequence according to STBC codes. The TDMA schedue designed by the sink wi decide which virtua MIMO ce transmits first and which the second, and so on. At the given schedue time sot, the sink wi request data transmission from the virtua MIMO ce. ISS : Vo. 4 o.06 June

3 3 WORK APPROACH In the anaysis, the foowing assumptions are made: i) There are nodes distributed uniformy in an M M region ii) An AWG (Additive White Gaussian oise) channe with squared path oss is assumed for the intracuster communication, iii) A fat Rayeigh fading channe with a power oss is assumed for the inter custer communication iv) BPSK is used as moduation scheme. 3.1 Communication Energy Consumption Mode In order to mode the energy consumption of the whoe network, the energy consumption of transmitting or receiving one bit is modeed firsty[8]. The tota average power consumption aong the signa path can be divided into two main components: the power consumption of a the power ampifiers P PA and the power consumption of a other circuit bocks P c. The first term P PA is dependent on the transmit power P out, which can be cacuated according.the transmitting energy consumption of one bit is defined as: E bt = (P PA + P C ) / R b Where E bt is the energy consumption of transmitting one bit when both circuitry and transmission energy consumption are considered, P PA is the power consumption of a power ampifiers, P C is the power consumption of a other circuit bocks and R b is the bit rate of the system. The power consumption of the power ampifiers can be approximated as : P PA = (1+ α) P out Where α is the efficiency of radio frequency power ampifier. Thus Pout can be estimated as: (4Π) M f EbRb d ; d d 0 GtGrλ Pout = M f 4 E > brb d ; d d o GrGtht hr Where E b is the required average energy per bit at the receiver for a given BER requirement Pb, R b is the system bit rate, d is the transmission distance, d 0 is the distance threshod and defined as 4πh t h r /λ. G t is the transmitter antenna gain, G r is the receiver antenna gain, λ is the carrier waveength, h t and h r are the heights of transmitter and receiver antenna, M is the ink margin compensating the hardware process variations and other additive background noise or interference. f is the receiver noise figure defined as f = r / 0, where 0 is the singesided therma noise power spectra density (PSD) at room temperature, and r is the PSD of the tota effective noise at the receiver input. Denote by P ct and P cr the power consumption of transmitting circuit bocks and receiving circuit bocks, respectivey. As in [11], P ct can be estimated as: P ct P DAC + P mix + P fit + P syn And P cr can be estimated as: P cr P LA + P syn + P mix +P IFA + P fir + P ADC where P DAC, P mix, P fit, P syn, P LA, P IFA, P fir, P ADC are the power consumption for D/A converter, the mixer, the fiters at the transmitter side, the frequency synthesizer, the ow noise ampifier, the intermediate frequency ampifier, the active fiters at the receiver side and the A/D converter, respectivey. The tota power consumption in a the circuit bocks during the ong-hau communications step consists of power consumption in T number of transmitter circuits and R number of receiver circuits: P C = T (P DAC +P mix +P fit +P syn ) + R (P LA +P syn +P mix +P IFA +P fir +P ADC ) I. Long hau transmission: During the ong-hau communication, the sensor node in a custer encodes the compressed data according to STBC scheme and transmits it to the access point or sink. The access point fies over the sensor fied to coect the data from sensor network. It retrieves the data of a custer when it is right above the custer. Since usuay the ong-hau distance between the access point and the custer is much arger than the maximum separation of the custer, we assume the ong-hau transmission distance, denoted by d tos, is the same for each transmitting node. The energy ISS : Vo. 4 o.06 June

4 consumption of transmitting one bit for ong hau MIMO transmission from custer head to sink can be cacuated as: E bt_mimo = (1+α) E b_mimo M G G h h r t f r t d 4 tos T P + R where d tos is the distance from the custer head to the sink, R b is the bit rate of the system, which is assumed to be equa to Bb, and B is the transmission bandwidth. E b_mimo is the required average energy per bit for a given BER II. Intra-custer transmission: In this phase, the non-custer head nodes send their data frames to the custer head during their aocated time sot. The duration and the number of frames are same for a custers and depend on the number of non-custer head nodes in the custer. The energy consumption of transmitting one bit from one custer member to custer head in one virtua MIMO ce can be cacuated as: E bt_intra = (1+α) E b_intra (4 ) M G G λ Π f dctoch t r b Pct + Bb Where E b_intra is the required energy per bit for a given BER requirement, d toch is the distance from the custer member to custer head in one virtua MIMO ce. III. Inter-custer transmission: After a custer head receives data frames from its custer members, it performs data aggregation and broadcasts the data to virtua MIMO sending nodes. When each cooperative sending node receives the data packet, they encode the data using space time bock code (STBC) and transmit the data cooperativey. The energy consumption of transmitting one bit from custer head to custer head in one virtua MIMO ce can be cacuated as: E bt_inter = (1+α) E b_inter (4 ) M G G λ Π f dctoc t r Pct + Bb where E b_inter is the required energy per bit for a given BER requirement, d CtoC is the distance from the custer head to custer head in one virtua MIMO ce. ct A. Tota Energy Consumption Mode of the Proposed Scheme Energy consumption of the proposed SCHCT scheme consists of two terms: energy consumption for the custer heads and energy consumption for the sensor nodes.[9] 1) Energy Consumption for custer heads: If there are K c custers, then there are average /K c nodes per custer. Then the tota energy required in one custer, E custer is given by: E custer = (/K c -1) E s +E CH Where E s is the energy consumption for a custer member and E CH is the energy consumption for a custer head. For each custer head, the energy consumption consists of receiving data from the custer members, aggregating the received data, transmitting the aggregated data to the cooperative custer heads, receiving aggregated data from other cooperative custer heads in the virtua MIMO ce, transmitting the encoded data to the sink by virtua MIMO technique. Therefore, the energy consumption for the custer head is given by: E CH =L(/K c -1)E br +L(/K c )E DA +L( T -1)E br +LE bt_inter +LE bt_mimo where E DA is the aggregation energy consumption per bit. ) Energy Consumption by sensor nodes: For sensor nodes in custers, the action is ony transmission of data to the custer head, so the energy consumption is given by: E s =LE bt_intra And the energy consumption for receiving one bit can be approximated by: ISS : Vo. 4 o.06 June

5 E br = P cr / (Bb) Based on the above anaysis, the overa energy consumption in one round of the SCHCT scheme can be derived as: E tota = K c E custer = (-K c ) E s + K c E CH The above equation gives the tota energy consumption in an M M MIMO network. By approximating the bound as equaity [10], we can cacuate the tota energy consumption per bit for the MISO system as: E MISO = (1+α) [ t o / P b 1/Mt ] [ (4πd) /G t G r λ ] [M f ] + (P c /R b ) 4 Simuation and resuts The anaysis of the proposed cooperative mutihop MIMO scheme is carried out using MATLAB to evauate the energy consumption and maximize the ifetime of the sensor network. Depoying the sensor nodes 50 static wireess sensor nodes wi be randomy depoyed in a 1000 x 1000 area; the communication range of each sensor node is 00 meters and the distance measurement range is assumed as the same as the communication range. A sensor nodes have radio modue for communication and are equipped with an utrasound transceiver for the distance measurement. It is aso assumed that each node has at east three one-hop neighbours. This is done to get unique reaization for the network scenario. According to the design scheme the sensor nodes are randomy depoyed in the given space then connecting each two nodes if the distance between them ess than or equa to the communication radius which is shown by the Figure-1.Figurewise description for,3,4,5,6,7. Distance d in m Distance d in m Figure 1(Depoyment of Sensor) Energy Consumption at different stages: A sensing fied of dimension M M (M =00 m) with a popuation of = 400 nodes is considered for simuation. Other system parameters are f c =.5GHz, B=10kHz, α=0.4706, M =40dB, f =10dB, G t G r =5dB, h t =h r =1m, 0 /=-174dBm/Hz, P ct =98.mW, P cr =11.6mW,P b =10-3,E DA =50nJ,L=1000 bits, K c =8, L=1000. ISS : Vo. 4 o.06 June

6 Figure :Energy Consumption per bit Figure 3:Energy Consumption in ong-hau communication Figure 4:For inter or intra custer communication ISS : Vo. 4 o.06 June

7 Figure 5:MISO v/s MIMO (4X4) Figure 6: Tota Energy Consumption over distance dtos SCHCT,t=8 SCHCT,T=4 LEACH Percentage of nodes aive Time(round) Figure 7: Percentage of nodes aive over time with distance to sink ISS : Vo. 4 o.06 June

8 5 Concusions Based on the basic mobie energy consumption mode, the overa energy consumption mode of the proposed scheme is derived which shows that athough the energy consumption in MISO is ess in comparison to other traditiona techniques but when distance from sink increases MIMO outperforms MISO.A custer-based cooperative MIMO scheme for mutihop WS has been expored to minimize the energy consumption and increase the ifetime of sensor nodes and the performance of the system is evauated. The energy consumption is ess when the number of transmitting antenna is minimized. When compared with LEACH scheme, numerica and simuation resuts together show that the proposed scheme can proong the sensor network ifetime greaty when the distance to sink is above a threshod, especiay in situations where the sink is far from the sensor area. REFERECES [1] V. Tarokh, H. Jafarkhani and A. Caderbank, Space- Time Bock Codes from orthogona Design, IEEE Trans. Inform. Theory, vo 45, no.5, pp , Juy [] S. Cui, A. J. Godsmith and A. Bahai, Energy efficiency of MIMO and Cooperative MIMO Techniques in Sensor etworks, IEEE Journa of Seected Areas in Communications, vo., pp , Aug. 004 [3] Yong Yuan, Min Chen and Taekyoung Kwon, A nove custer-based cooperative MIMO scheme for muti-hop wireess sensor networks, EURASIP Journa on Wireess Communications and etworking, vo. 006, pp. 1-9, 006. [4] Y. Yuan, Z. He and M. Chen, Virtua MIMO- based cross-ayer design for wireess sensor networks, IEEE Transactions on Vehicuar Technoogy, vo. 55, no.3, pp , 006 [5] X. Li, M. Chen and W. Liu, "Appication of STBC encoded cooperative transmissions in wireess sensor networks", IEEE Signa V. Tarokh, H. Jafarkham, A. R. Caderbank, Space-Time Bock Coding for Wireess [6] Communications: Performance Resuts, IEEE Journa on Seected Areas in Communications. Vo. 17, o. 3, pp , March 1999Processing Letters, Vo., o., pp , February 005. [7] S. Bandyopadhyay and E. J. Coye, An energy efficient hierarchica custering agorithm for wireess sensor networks, in Proceedings of IFOCOM 003, Apri 003. [8] Cross ayer optimization in energy constrained networks by Aung Aung and Shuguang Cui, Stanford University [9] Yong Yuan, Min Chen and Taekyoung Kwon, A nove custer-based cooperative MIMO scheme for muti-hop wireess sensor networks, EURASIP Journa on Wireess Communications and etworking, vo. 006, pp. 1-9, 006. [10] Sajid Hussain, Anwaru Azim,Jong Hyuk Park, An Energy-Efficient Virtua MIMO Transmission Scheme for Custer-based Wireess Sensor etworks ISS : Vo. 4 o.06 June