TVWS Power Line Communication System for Indoor Networks

Transcription

1 TVWS Power Line Communication System for Indoor Networks Mohammad Heggo, Xu Zhu, Yi Huang Department of Electrical Engineering and Electronics The University of Liverpool Liverpool, L9 3GJ, UK s: {mheggo, xuzhu, Sumei Sun Institute for Infocomm Research Agency for Science, Engineering and Research, Singapore Abstract Broadband power line communication (BPLC) is a promising technology for satisfying the growing data rate demands in the indoor networks However, the BPLC transmission power is restricted in the VHF band to avoid harmful interference to the existing wireless services In this paper, a new cooperative system is proposed between the BPLC and the cognitive wireless communication over the TV white space (TVWS) We focus on maximizing the VHF capacity of the BPLC The proposed system improves the achievable ergodic capacity Also, a new iterative precoding algorithm is proposed in order to satisfy the interference limit at the primary user (PU)and enhance the ergodic capacity Finally, the simulations demonstrate the significant enhancement in the achieved capacity for the BPLC in the VHF band I INTRODUCTION Broadband power line communication has drawn the attention of the researchers in the last decade as it offers the indoor networks a high speed inexpensive solution [1] However, the electromagnetic compatibility (EMC) with wireless services remains a crucial problem for the BPLC transceivers Hence, in the IEEE 191 standard the BPLC transmission power in the VHF band is restricted so as to satisfy the limit of power spectral density (PSD) of -5 dbm/hz [] The latter constraint is significantly limiting the capacity of the BPLC in the VHF band and hence is considered as a crucial problem for the BPLC Enhancing the BPLC capacity and satisfying the interference limit with wireless devices had been approached in the previous literature using two methods: (1) Cognitive BPLC as in [3] [] and () Hybrid wireless BPLC as in [7] For the cognitive BPLC proposed in [3] [], the BPLC transmitter is continuously sensing the spectrum before accessing using the coupling circuit receiver This procedure avoides harmful interference with the neighbouring wireless devices working in the same band Also, the procedure mitigates the narrow band interference noise from the wireless devices However the main drawback in the cognitive BPLC is that it consideres the power line cables as very good antennas for all the interference VHF band with the wireless services This assumption is impractical since it neglects the variation in the near field coupling between the BPLC and the VHF wireless signals which is dependent on the frequency as previously addressed in [] Hence, at some frequencies the coupling is insufficient for the detection of the primary user (PU) presence which can lead to false access to the spectrum Also, weak coupling between the power line signals and the wireless VHF band can lead to a long sensing time that yields a capacity loss It is worth mentioning that according to [9] it is requested for the secondary user to detect the presence of the TV or radio signals at very low power levels of -1 dbm This restriction puts an obstacle in front of applying the cognitive BPLC On the other side for the hybrid wireless BPLC, Wi-Fi was proposed in [7] to enhance the capacity of the BPLC However, the solution becomes inapplicable in the VHF band This is due to the transmission power constraint of -5 dbm/hz set by the IEEE 191 standard This limit will allocate most of the power to the Wi-Fi channel and the overall system will turn to be a Wi-Fi transceiver In this paper, a new cooperative system between the BPLC and the TVWS is proposed Our contribution in the proposed system can be addressed as follows: First, the proposed system improves significantly the VHF BPLC capacity through adding the wireless TVWS channel and using the advantage of the multiple-input multiple-output (MIMO) channel Second, the interference with the wireless services is mitigated using a new proposed iterative precoding technique at the transmitter The proposed algorithm takes the advantage of the multipleinput single-output (MISO) channel between the TVWS- BPLC transmitter and the wireless receiver Third, compared to the cognitive BPLC our proposed system offers stable cognitive spectrum sensing independent of the coupling level between the BPLC and the wireless media Fourth, compared to the hybrid Wi-Fi BPLC, our system offers an inexpensive hybrid system This is due to the elimination of the RF upand down-converters needed for the Wi-Fi transceiver Fifth, the TVWS offers lower channel path loss than the Wi-Fi bands in the GHz and 5GHz [1] [11] The paper is organized as follows: In Section II, the proposed system model including the capacity and channel models is presented In Section III, the cognitive spectrum access is addressed Our iterative precoding technique and the MIMO spectrum sensing are presented Also, the power allocation for the MIMO channel is presented In Section IV, the capacity simulation results of our proposed system are compared with the conventional cognitive BPLC Finally, in

2 ) The channel is idle and detected occupied r 1 = Σ nr i=1 log (1 +,ip 1,i ) () σ n 3) The channel is occupied and detected idle r 1 = Σ nr i=1 log (1 + λ,ip,i σn + σp ) (3) ) The channel is occupied and detected occupied Fig 1 System Model Section V the paper is concluded Notations: E{} denotes the expectation operator, x + denotes max(, x), the vectors are represented by boldface letters and the conjugate transpose of a matrix or a vector A is A I denotes the identity matrix II SYSTEM MODEL A MIMO TVWS-BPLC Model The proposed TVWS-BPLC system model is illustrated in Fig 1 Complete channel state information (CSI) is assumed at the transmitter The binary encoded data is modulated using QAM modulator Each QAM symbol is pre-coded using either direct singular value decomposition (D-SVD) algorithm or projected singular value decomposition (P-SVD) algorithm The precoding algorithm is discussed in detail in Section III At the receiver, the received signal is decoded using a decoding matrix according to the pre-coding type The decoded data is further mapped to a QAM demodulator to get the binary information Since the proposed system is accessing the TVWS, a cognitive spectrum access algorithm is proposed In the algorithm the channel between a single secondary user (SU) transmitter and receiver is considered to be n r n r MIMO channel where n r is the number of SU channel links However, the channel between an SU transmitter and a single primary user (PU) receiver can be considered as n r 1 MISO channel This advantage is used by our proposed system for better SU to PU interference mitigation B Capacity Model In [], a system model is proposed for the SU cognitive access In this model the SU is accessing the cognitive spectrum with two power levels P and P 1 in the PU absence and presence respectively The system proposed in [] was designed for SISO channel between secondary users Also the interference channel between SU and PU is SISO In our work, we develop the model proposed in [] to represent the MIMO channel between secondary users Also, the model is modified to represent the MISO interference channel between SU and PU Hence, the four MIMO capacity levels can be modelled as follows: 1) The channel is idle and detected idle r = Σ nr i=1 log (1 + λ,ip,i ) (1) σ n r 11 = Σ nr i=1 log (1 +,ip 1,i σn + σp ) () where σn and σp are the noise power and the primary user power respectively P,i and P 1,i are the SU signal power of the i-th channel link in case of the PU absence and presence respectively λ,i and,i represent the singular values of i- th linknel link of the idle and occupied channel respectively Hence, we can conclude the MIMO ergodic capacity as follows: C = ( T τ ) EH,g {P (H )(1 P fa )r + P (H )P fa r 1 T + P (H 1 )(1 P d )r 1 + P (H 1 )P d r 11 } (5) where P (H ) and P (H 1 ) are the probabilities of PU absence and presence respectively P d and P fa are the PU detection probability and false alarm probability respectively T is the symbol time duration and τ is the PU detection time duration H represents the MIMO channel gain matrix between the SU transmitter and the SU receiver while g represents the MISO channel gain vector between the SU transmitter and the PU receiver For a constrained power communication system, the average MIMO power of the SU shall be less than a predefined value P av as follows: E H,g {P (H )(1 P fa )Σ nr i=1 P,i + P (H )P fa Σ nr i=1 P 1,i +P (H 1 )(1 P d )Σ nr i=1 P,i + P (H 1 )P d Σ nr i=1 P 1,i} P av () Also, the MISO interference to the PU is limited to a certain value Γ as follows: E H,g {P (H 1 )(1 P d ) g U P +P (H 1 )P d g 1 U 1 P 1 } Γ U and U 1 are the pre-coding matrices in case of PU absence and PU presence respectively Also g and g 1 are the selected SU-PU channel gain for the cases of PU absence and presence respectively The criterion of selection is according to the precoding scheme P and P 1 are the SU signal power vectors in case of PU absence and presence respectively where P,i and P 1,i are their i-th elements respectively III COGNITIVE SPECTRUM ACCESS A Iterative Hybrid D-SVD/P-SVD Precoding Technique In our work, a new algorithm is proposed for cognitive spectrum access It is assumed that CSI of the PU receiver is known at the SU transmitter The algorithm pre-codes the SU (7)

3 transmitter data according to the SU-PU channel Two precoding algorithms are being used: (1) Direct singular value decomposition (D-SVD) () Projected singular value decomposition (P-SVD) In case of D-SVD pre-coding, let H be the n r n r channel matrix between the TVWS-PLC transceivers H is decomposed using D-SVD into H = QΛ 1/ U Q and U are n r n r matrices with orthonormal columns Λ is an n r n r diagonal positive matrix with λ vector as its diagonal Let V = U is the pre-coding matrix However, in the P-SVD pre-coding, let ĝ 1 (n) = g 1 (n)/ g 1 (n) be the unit vector in the direction of g 1 which represents the channel vector between the SU and the PU We can define the projection of the channel matrix H in the null space of g 1 to be H Then using the SVD of H = Q Λ 1/ U we can conclude the precoding matrix V = U and decode the received signal by multiplying by Q In our proposed pre-coding algorithm, both D-SVD and P-SVD are used jointly according to the CSI 1) Idle Channel: In case of idle channel, D-SVD is used as pre-coding scheme Although the primary user is assumed not to be accessing the channel, an interference limit is forced for false detection probability 1 P d to avoid harmful interference The SU-PU channel g,k which satisfies max nt g,k U is selected for achieving the interference limit in (7) where k = 1,, n t and n t is the number of primary users In other words, the interference channel with maximum gain is selected as the worst condition of interference with the primary user ) Occupied Channel: In this case, the primary user is assumed to be accessing the channel In our proposed algorithm the D-SVD is also used as a default pre-coding scheme as in the case of the idle channel However, the decision is taken to switch the pre-coding scheme to P-SVD according to the CSI The decision is taken to satisfy two conditions: (1) Achieve the interference limit at the PU receiver () Achieve high SNR at the SU receiver Hence, the previous two conditions can be satisfied knowing the CSI as follow From equation (7) it can be concluded that: Σ nr i=1 α Γ i,kp i,k () P (H 1 )P d where α i,k is the i-th component of the vector α k = g 1,k U 1 of the k-th primary user In order to have good SNR at the receiver we shall have: ΛP 1 σ n + σ p (9) It can be assumed that the level of interference at the PU Γ σ n to avoid harmful interference Also we can assume that σ n + σ p = Γ where is arbitary constant that satisfies 1 Hence, the following equation can be concluded: ΛP 1 α k P 1 (1) This can be translated in the following matrix form: α 1,k α nr,k P 1 α nr,k α 1,k P nr (11) Directly, it can be concluded det α 1,k α nr,k α nr,k α 1,k () where λ α k The condition in () shall be satisfied for the use of D-SVD If this condition is violated, then P-SVD is used to eliminate one primary user The eliminated primary user is selected to satisfy max nt { α k } Hence, the algorithm proposed for cognitive access can be summarized as follows: Proposed iterative hybrid D-SVD/P-SVD algorithm Initialize the pre-coding algorithm as: D-SVD and n = n r Repeat: Calculate the determinant of matrix in () and check the condition If the condition satisfies end the loop, otherwise change the pre-coding to P-SVD n = n 1 If n >, Select the SU-PU interference channel to be cancelled that satisfies max nt { α k and go to Repeat If n =, opportunistic access is stopped end B Cognitive Sensing It is known that the sensing time is considered as a challenging issue for the cognitive systems In our proposed system we derive the formula for the optimum sensing time for a MIMO cognitive system Since we are using TVWS-PLC MIMO receiver for cognitive sensing, the probability of PU detection using OR rule for energy detection method can be derived as follows: P d = 1 Π nr i=1 (1 P d i ) (13) where P di is the probability of detection for i-th link, it can be expressed as follows: P di = Q ( ( ɛ τ i f s σn γ i 1)( γ i + 1 )) () Where ɛ is the detection threshold, γ i is the received PU signal to noise ratio, f s is the sampling frequency and Q() is the complementary error function Similarly, we can express both the MIMO and single channel link false alarm probabilities as in [13]: P fa = 1 Π nr i=1 (1 P fa i ) (15) P fai = Q ( ( ɛ σn 1)( τ i f s ) ) (1)

4 Let N mini = τ i f s be the minimum number of samples requested to achieve the detection and false alarm probabilities for i-th channel linkn mini can be expressed as follows: N mini = 1 γi (Q 1 (P fai ) Q 1 (P di ) γ i + 1) (17) We can assume that the requested false alarm probability of all channel links are equal ˆPfa Also, the detection time and the number of samples are the same for all channel links Hence, relation between the detection probabilities of two links i and m can be expressed as follows: γ m γ i P di = Q( γ m γi + 1 Q 1 ( ˆP fa ) + Q 1 γm + 1 (P dm ) γ i + 1 ) (1) For i, m = 1, n r, let the total MIMO P d in (13) is requested to achieve a predefined value ˆP d Hence, P dm of each m-th channel link can be concluded as follows: ( γ Π nr m γ i i=1 1 Q( γ m γi + 1 Q 1 ( ˆP fa ) +Q 1 γm + 1 ) (19) (P dm ) = 1 γ i + 1 ˆP d This equation can be solved numerically for any P dm, then using P dm we can get the rest of P di )in equation (19) Also, N mini and τ mini can be obtained from equation (17) directly A Simulation Setup IV SIMULATION In this section, the simulation results are presented for our proposed system compared to the conventional cognitive BPLC The OFDM symbol duration T is taken to be 5 ms The channel bandwidth is taken as MHz and the frequency of operation is assumed to be 1 MHz The requested probability of false alarm and detection probability are 1 7 and 9999 respectively Also, the probability of PU presence is The received primary user SNR is assumed to be - db n r and n t are assumed to be equal For the simulation of the power line environment, we use the random topology generator presented in [] Each iteration a random power line topology is generated and points are randomly selected to represent secondary users and primary users For computing the PLC path loss, the electric distance between the transmitter and receiver is computed (ie the total length of the separating cables) However for the wireless path loss the geometric distance is calculated (ie the difference in the Cartesian coordinates) B Simulation Results 1) Ergodic capacity at different transmission power levels: In Fig the average ergodic capacity of our proposed system is compared to conventional cognitive BPLC system for different levels of transmission power Two remarkable observations can be addressed: First, our proposed system significantly improves the ergodic capacity compared to the conventional cognitive BPLC Second, the increase in transmission power Capacity (Bits/Sec/Hz) TVWS PLC Capacity at Pav = db TVWS PLC Capacity at Pav = 5 db TVWS PLC Capacity at Pav = 1 db Cognitive PLC Capacity at Pav = db Cognitive PLC Capacity at Pav = 5 db Cognitive PLC Capacity at Pav = 1 db SU SU Distance (m) Fig Average TVWS PLC ergodic capacity vs the distance between the SU transmitter and the SU receiver at different transmission power levels Capacity (Bits/Sec/Hz) TVWS PLC Capacity at Pav = db TVWS PLC Capacity at Pav = 5 db TVWS PLC Capacity at Pav = 1 db Cognitive PLC Capacity at Pav = db Cognitive PLC Capacity at Pav = 5 db Cognitive PLC Capacity at Pav = 1 db PU SU to SU SU Distance Ratio Fig 3 Average TVWS PLC ergodic capacity vs the ratio between the PU Rx-SU Tx distance and the SU Rx-SU Tx distance at different transmission power levels yields a considerable increase in the ergodic capacity for the TVWS-BPLC system However, for the cognitive BPLC the increase in the transmission power leads to slight increase in the capacity In Fig 3 it is shown that our proposed TVWS-BPLC system capacity saturates rapidly as the distance ratio increases It can be observed that the capacity obtained by the conventional cognitive BPLC at distance ratio equals to one is obtained by our proposed model at significantly lower distance ratio This proves that our proposed system is much less interfering to the PU receiver even if it is located at very close distance to the TVWS-BPLC transmitter Also, it can be observed that increasing the level of the transmission power yields a corresponding increase in the system capacity regardless of the PU receiver separation from the SU transmitter This is due to the use of our proposed iterative precoding algorithm that has the ability to mitigate the interference with the nearby PU receiver ) Ergodic capacity at different coupling loss levels: The coupling loss represents the coupling between the BPLC signal and the VHF wireless band As the coupling increases the coupling loss decreases and vice versa In Fig the effect of the coupling loss on the ergodic

5 Capacity (Bits/Sec/Hz) TVWS PLC Capacity at Coupling Loss = 1 db TVWS PLC Capacity at Coupling Loss = db TVWS PLC Capacity at Coupling Loss = 5 db Cognitive PLC Capacity at Coupling Loss = 1 db Cognitive PLC Capacity at Coupling Loss = db Cognitive PLC Capacity at Coupling Loss = 5 db SU SU Distance (m) Fig Average TVWS PLC ergodic capacity vs the distance between the SU transmitter and the SU receiver at different coupling loss levels Sensing Time (ms) TVWS PLC sensing time Cognitive PLC sensing time 1 Coupling Loss (db) Fig 5 Spectrum sensing time vs coupling loss capacity is modelled It can be shown for the TVWS-BPLC system that as the coupling loss increases a slight increase in the ergodic capacity is observed This can be understood as the coupling loss increases the BPLC interference to the PU decreases which yields the allocation of more power to the BPLC Hence, the overall capacity increases as the BPLC has a better path loss over the wireless VHF On the other side, for the cognitive BPLC as the coupling loss increases the ergodic capacity decreases This is due to the corresponding increase in the sensing time The effect of the sensing time increase is much more than the decrease in the interference of the BPLC to the VHF band mentioned above Hence, it can be concluded that our proposed system preserves the capacity performance regardless of the coupling loss In Fig 5, it can be shown the effect of the coupling loss increase on increasing the sensing time of the conventional cognitive BPLC However, for our proposed system since we are using MIMO sensing scheme, the increase in the coupling loss is not affecting the overall sensing time V CONCLUSION In this paper, the TVWS-BPLC system has been proposed for the indoor communication networks The proposed system offers an inexpensive solution for enhancing the BPLC capacity in the VHF band Also, an iterative precoding algorithm has been proposed to mitigate the interference with the PU receivers in the VHF band Through simulations it has been shown that our proposed system improves the BPLC capacity significantly We have also shown that the capacity enhancement has been preserved under small separation distances from the PU receiver Moreover, the proposed system demonstrates robust performance against the variation in the BPLC coupling loss which affects the cognitive BPLC sensing time and capacity in the VHF band REFERENCES [1] M U Rehman, S Wang, Y Liu, S Chen, X Chen, and C G Parini, Achieving high data rate in multiband-ofdm UWB over power-line communication system, IEEE Transactions on Power Delivery, vol 7, no 3, pp , July [] I S Association et al, IEEE Standard for Broadband over Power Line Networks: Medium Access Control and Physical Layer Specifications, IEEE Std 191, vol 1, pp 1 15, 1 [3] B Praho, M Tlich, P Pagani, A Zeddam, and F Nouvel, Cognitive detection method of radio frequencies on power line networks, in Proc IEEE International Symposium on Power Line Communications and Its Applications (ISPLC), pp 5 3, March 1, Rio de Janeiro Brazil [] R Vuohtoniemi, J-P Makela, J Vartiainen, and J Iinatti, Detection of broadcast signals in cognitive radio based PLC using the FCME algorithm, in Proc IEEE International Symposium on Power Line Communications and its Applications (ISPLC), pp 7 7, March, Glasgow Scotland [5] K M Ali, G G Messier, and S W Lai, DSL and PLC co-existence: An interference cancellation approach, IEEE Transactions on Communications, vol, no 9, pp , September [] S W Oh, Y L Chiu, K N Ng, R Mo, Y Ma, Y Zeng, and A Syed Naveen, Cognitive power line communication system for multiple channel access, in IEEE International Symposium on Power Line Communications and Its Applications, pp 7 5, March 9, Dresden Germany [7] S W Lai, N Shabehpour, G G Messier, and L Lampe, Performance of wireless/power line media diversity in the office environment, in Proc IEEE Global Communications Conference (GLOBECOM), pp 97 97, December, Texas USA [] M Karduri, M D Cox, and N J Champagne, Near-field coupling between broadband over power line (BPL) and high-frequency communication systems, IEEE Transactions on Power Delivery, vol 1, no, pp , October [9] Second Report and Order and Memorandum Opinion and Order In the Matter of Unlicensed Operation in the TV Broadcast Bands, Additional Spectrum for Unlicensed Devices Below 9 MHz and in the 3 GHz Band, Federal Communication Commision, Document -, November [1] A H Ali, M A Razak, M Hidayab, S A Azman, M Z M Jasmin, and M A Zainol, Investigation of indoor WIFI radio signal propagation, in Proc IEEE Symposium on Industrial Electronics & Applications (ISIEA) 1, pp , October 1, Penang Malaysia [11] J Andrusenko, R L Miller, J A Abrahamson, N M Merheb Emanuelli, R S Pattay, and R M Shuford, VHF general urban path loss model for short range ground-to-ground communications, IEEE Transactions on Antennas and Propagation, vol 5, no 1, pp , October [] S Stotas and A Nallanathan, Enhancing the capacity of spectrum sharing cognitive radio networks, IEEE Transactions on Vehicular Technology, vol, no, pp , October 11 [13] Y-C Liang, Y Zeng, E C Peh, and A T Hoang, Sensing-throughput tradeoff for cognitive radio networks, IEEE Transactions on Wireless Communications, vol 7, no, pp , April [] A M Tonello and F Versolatto, Bottom-up statistical PLC channel modelingpart I: Random topology model and efficient transfer function computation, IEEE Transactions on Power Delivery, vol, no, pp 91 9, April 11