Coherent Detection Method with Compensation at Transmitter in Time Division Duplex System

Transcription

1 Coherent Detetion Method with Compensation at Transmitter in Time Division Duplex System Young An Kim 1, Choong Seon Hong 1 1 Department o Eletronis and Inormation, Kyung Hee University, 1 Seoheon, Giheung, Yongin, Gyeonggi, Korea, roundsun@networking.khu.a.kr, shong@khu.a.kr Abstrat We proposed a pre-ompensation method ombining two wireless ommuniation shemes, TDD(Time Division Duplex) whose orward and reverse links use the same requeny, and the oherent detetion as the best detetion sheme to give a better mobile to base stations type ommuniation system. The method is that the mobile station observes and estimates the hannel harateristi rom the reeived signal, and pre-ompensates the transmitted signal adapting to the ading hannel to improve the perormane. Although preamble signals are neessary in the onventional method to observe and estimate the hannel harateristi, the proposed method needs not the preamble signals, whih derease the inormation rate. Computer simulations show better BER perormane than the dierential detetion that is usually used in Rayleigh ading hannel. 1. Introdution Reently, the growing demand o mobile ommuniation like portable telephones, ar telephones and PHS (Personal Handy phone System) inreases rapidly [1]. We need ommuniation systems with more apaity and higher quality. One o suitable or the demand is Time Division Duplex (TDD) system [2-5], in whih only one requeny hannel is used or transmitting and reeiving signals between the base and mobile stations. TDD sheme is eetively applied to pre-diversity or power ontrol in mobile ommuniation [2-5]. The other one is oherent detetion with advantages that less signal-to-noiseratio (SNR) than dierential detetion on the same Biterror-rate (BER) is required. However, it is diiult to arry out the oherent detetion due to hanging amplitude and phase in the ading hannel. Thereore preamble signals are neessary to estimate the hannel harateristi and ompensate it at the onventional oherent detetor [6-7]. In this paper, we propose a new oherent detetion method or mobile radio ommuniation using TDD sheme. The method is that the mobile station observes and estimates the harateristis o the hannel rom the reeived signal transmitted at base station on orward link, and ompensates the ading in advane. Although preamble signals are neessary in onventional methods, the method we proposed needs not the preamble signals whih derease the inormation rate. Computer simulations show better BER perormane than the dierential detetion, whih has been usually used in Rayleigh ading hannel. The paper is organized as ollows. TDD system and the oherent detetion explained in hapter 2. The proposed system is desribed in hapter 3. Computer simulation results are presented and disussed in hapter4, and inally, the onlusions are disussed in hapter5. 2. Conventional System 2.1 Coherent detetion Fig. 1 shows the typial oherent detetion system or BPSK(Binary Phase Shit Keying) modulation in the transmitter and reeiver with the osillator 1 The work was supported by the Korea Researh Foundation Grant unded by the Korean Government (MOEHRD)" (KRF D00394) 2007 International Conerene on Multimedia and Ubiquitous Engineering(MUE'07)

2 requenies and requenies must be same as ' respetively. These ' and the phases = between the reeived signal and loal osillator at the reeiver must be adjusted or oherent detetion. It is diiult or the oherent detetor to detet a signal in the ading hannel, sine the phase and amplitude lutuations disturbs the detetion system. Thereore, we need preamble signals that estimate the ading hannel harateristis or stabilizing the system in the onventional oherent detetion. However, the preamble signals embedded in the transmitted signal redue the inormation rate. 2.2 TDD System Fig. 1 Coherent Detetion The FDD(Frequeny Division Duplex) tehnique has traditionally been used or both wireless and wired duplex ommuniations. In the FDD systems two separate requeny hannels are used or transmitting and reeiving signals as illustrated in Fig. 2. Fig. 3 TDD System The TDD tehnique(also known as Ping Pong) has the advantages as ollows. (i). Guard band, whih separates orward and reverse hannels, is unneessary. Thereore it is easy to requeny bands. (ii). The system generally oers an eonomial system design. Sine both the orward and the reverse links use the same requeny hannel, the reeiver and the transmitter ilters and osillators an be the same. (iii). The bandwidth o TDD system is twie wider than FDD system. Filters in TDD system do not need to have higher Q as FDD system does. (iv). The utilization o the same requeny band or both orward and reverse hannels results in high orrelation between the requeny patterns o the orward and reverse links. There have been a lot o reports using TDD advantages. In this paper, we utilize (iv) mainly. The onept o whole TDD system is shown in Fig. 4. Fig. 2 FDD System In the TDD system the orward and the reverse link ommuniations are arried out in the same requeny band, with the vase station and the mobile station alternately transmitting and reeiving inormation signals as illustrated in Fig. 3. Frame and Burst are deined as Fig. 3. Fig. 4 TDD System 2007 International Conerene on Multimedia and Ubiquitous Engineering(MUE'07)

3 3. Proposed System In the proposed oherent detetion system, the phase lutuation in the ading and the requeny oset between the reeived signal and loal osillators have to be ompensated using TDD sheme. The method or the estimation and ompensation are explained as ollows. 3.1 Estimation method o requeny oset There have been some reports about estimating the requeny oset in blok demodulators. The proposed system estimate the requeny oset by the ollowing method. Multiplying the reeived signal and loal osillator at the reeiver with the ixed requeny oset between the transmitter and reeiver that have to be estimated and ompensated demodulates the data. The phase lutuating values due to the requeny oset is muh larger than one due to the ading. So it is possible to get an approximate value by alulating an average or a long period and many samples. The requeny oset is ound by alulation as ollowing equations k 1 ( θn+ 1 θn)/( k 1) n= 0 = (1) Where k and θ are the total sample number (the total n bits) in one rame and the phase at the reeived bit whih is removed rom the data omponent respetively. The rames are shown in Fig. 3. θ n an be written as; 1 2θ artan( Rθ I θn = 2 2 ) (2) 2 θ θ Where θ and θ are the real part and imaginary part o θ n. R I 3.2 Pre-ompensation method o phase lutuations Here we desribe a pre-ompensation method to stabilize the oherent detetor at the base station. Using the values o (1), (2) the phase lutuation is pre-ompensated. The transmitting signal rom the base station is expressed as ollows. Aos(2 π t+ θ ), { 0( data = θ = 0) (3) d d R I π ( data= 1) Where A, and θd are the amplitude o signals, the requeny o the base station osillator and the binary modulated phase. Demodulated signal at mobile station ignoring the noise omponent is A' os( θd + θ 2 π t ) (4) 2 Where A is proportional to A, is the requeny oset between base station osillator and mobile station osillator, and θ is the phase lutuation due to the ading. The transmitted signal rom the mobile station and the demodulated signal without any ompensation method ignoring the noise omponent are ollows. Bos{2 ( + ) t+ } (5) π θ d B ' os( θd + θ + 2 π t ) (6) 2 In use o the ompensation method the transmitted signal rom the mobile station and demodulated signal ignoring the noise omponent are written as; B os{2 π( + ) t+ θ θ 2 π t } (7) Where B 2 d ' os θd ( θ θ, ) = = (8) θ is the estimated phase lutuation and is the estimated requeny oset. Seeing (8), we understand possibility to ompensate the ading eet. 3.3 Demodulation method at base station At the base station, the error o ompensated phase lutuation at the mobile station due to ading is traked and revised by LMS algorithm deined as ollows. Y = C V (9) m m 1 m em dm Y m = (10) C = C + µ e Vm (11) Where V, C, Y, µ, d and are the input signal * m m 1 m m m m m e m vetor, oeiient vetor, the output value, step gain, ideal value o input and oeiient o error * respetively. V is the onjugate o V. m The error o the pre-ompensated requeny oset at the mobile station is revised as ollows. The error at eah rame(the phase is assumed to be onstant in the m 2007 International Conerene on Multimedia and Ubiquitous Engineering(MUE'07)

4 rame) is derived by the error o the requeny oset pre-ompensation at the mobile station. And the values o the error are adding up every rame, thereore signals are demodulated subtrating the total value o error. Proposed base station system is shown in Fig. 5. Fig. 6 BER perormane versus step gain 4.2 BER perormane ompared with onventional method Fig. 5 Base station system in TDD 4. Simulation Results Table.1 shows the onditions or omputer simulations. The ading has Rayleigh distributed amplitude and uniormly distributed phase rom zero to 2π. Table 1. Simulation Conditions Modulation Data rate Frame length Frequeny oset Doppler requeny Number o rame to estimate requeny oset Coherent BPSK 64 [kbit/se] 100 [bit] 1000 [Hz] 10 [Hz] 100 [rame] Fig. 7 shows the BER perormane versus SNR on the base station in the proposed system, where the Doppler requeny is 10 Hz. The proposed system ahieves a signiiant improvement that is about 4 db ompared with DPSK system at BER=10-3. The reason is that DPSK system is inluened by the phase lutuation due to the requeny oset and ading, but the proposed system estimates and ompensates these inluenes. 4.1 Parameter deision o LMS algorithm Fig. 6 shows BER perormane versus step gain where Doppler requeny and SNR are 10 Hz and 5dB respetively. BER is minimum at about 0.04 as shown in Fig. 6. The parameter the step gain o LMS algorithm should be set at Fig. 7 BER perormane vs. SNR at base station o proposed system However there is a loor at BER perormane, beause o the existene o rapid phase lutuations, whih annot be ompensated at the transmitter and reeiver International Conerene on Multimedia and Ubiquitous Engineering(MUE'07)

5 Fig. 8 shows BER perormane versus SNR in the proposed system and onventional system. The onventional ompensation method by LMS algorithm uses already known signals (10 bits) in every rame and is a post-ompensation. As Fig. 8 shows, there is also a loor o BER perormane in using LMS algorithm method. The loor an be shited down to less than BER = 10-3 and whih is enough or voie ommuniation by interleave and orward error orretion. And so on it is possible to improve the loor. Fig. 9 shows BER perormane versus SNR on the base station in the proposed system, whose parameter is the rame length. The rame length is shorter; the BER perormane is better perormane. The reason is that the rame length is shorter; the auto orrelation between the orward and reverse links in TDD is higher. Thereore the phase lutuation is estimated and ompensated well. 4.4 BER perormane on variety o Doppler requenies Fig. 8 BER perormane vs. SNR 4.3 BER perormane on variety o rame length Fig. 10 BER perormane vs. SNR on base station o proposed system Fig. 10 shows BER perormane versus SNR on the base station in the proposed system whose parameter is the Doppler requeny. The Doppler requeny is smaller; the BER perormane is better perormane. The reason is that the Doppler requeny is smaller; the phase lutuation in one rame is smaller. Thus, the auto orrelation between the orward and reverse links is higher; the phase lutuation is estimated and ompensated well. 5. Conlusion Fig. 9 BER perormane vs. SNR on base station o proposed system In this paper, pre-ompensation method or a new oherent detetion using TDD sheme has been proposed. Our proposed sheme is that ater the ading hannel harateristi is estimated, the transmit signal is ompensated at the transmitter and revised at the reeiver. This sheme does not need known data used in onventional systems International Conerene on Multimedia and Ubiquitous Engineering(MUE'07)

6 The omputer simulation showed that the proposed system ahieves about 4dB improvement ompared with DPSK system usually used in the mobile ommuniations at BER = Comparing with the onventional oherent system with known data (preamble data), the proposed system ahieved the improvement o transmitting eiieny and almost the same BER perormane. Reerenes [1] Masao Nakagawa: Personal Communiation and Consumer Communiation. Baihuukan. [2] Young-an KIM, Choong-seon HONG: R-ALOHA Protool or Inter-Vehile Communiation in Ad-ho Network. Korea ITS Conerene 2002, (2002), pp [3] Kazuhiko Fukawa, Hiroshi SUZUKI: An Adaptive Signal Proessing without Inormation o Intererene Signal. TECHNICAL REPORT OF IEICE. RCS04-27, (2004), pp [4] H. Singh, S. Singh: A MAC Protool based on Adaptive Beaon Forming or Ad Ho Networks. In Proeeding o the 14 th IEEE Personal, Indoor and Mobile Radio Communiation Conerene, (2003). [5] Akinori Hirukawa, Hitoshi Takanashi: Inter-Base-Station TDMA Frame Synhronization Tehnique or Street Miroellular System. IEEE PIMRC 94.,Vol. D-3, (1994), pp [6] Takeshi CHISHAKI, Nobuak INOUE: Engineering o Trai Planning. Kyoritsu Syuppan, Japan, (1993). [7] James. A. Roberts, Juan M.Bargallo: DPSK Perormane or Indoor wireless Riian Fading Channels. IEEE Trans. Commun., Vol om-42, (1994), pp International Conerene on Multimedia and Ubiquitous Engineering(MUE'07)