Simulation Model for a Frequency-Selective Land Mobile Satellite Communication Channel

Transcription

1 Vo 3, No., Simuation Mode for a Frequency-Seective Land Mobie Sateite Communication Channe Abstract Zachaeus K. Adeyemo, Oumide O. Ajayi *, Festus K. Ojo 3. Department of Eectronic and Eectrica Engineering, Ladoke Akintoa University of Technoogy, PMB 4, Ogbomoso, Nigeria. Department of Eectronic and Eectrica Engineering, Ladoke Akintoa University of Technoogy, PMB 4, Ogbomoso, Nigeria 3. Department of Eectronic and Eectrica Engineering, Ladoke Akintoa University of Technoogy, PMB 4, Ogbomoso, Nigeria * E-mai of corresponding author: skifutech@gmai.com This paper investigates a three-state simuation mode for a frequency-seective and mobie sateite communication (LMSC) channe. Aside from ionospheric effects, the propagation channes for LMSC systems are aso characterized by wideband effects due to mutipath fading which makes the channes time-variant and exhibit frequency-seective distortion. Hence, an adequate knowedge and modeing of the propagation channe is necessary for the design and performance evauation of the LMSC systems. A three-state simuation mode for a frequency-seective LMSC channe, which is a combination of Rayeigh, Rician and Loo fading processes, is deveoped. The propagation characteristics of the proposed LMSC channe mode are presented, and comparisons are made with the Rayeigh, Rician and Loo fading channes using bit error rate (BER) as the figure of merit. The simuation resuts show that the degree of fading experienced by the LMSC ink depends on the ength of time the mobie termina is in a particuar state or ocation, depending on the assumed probabiity of occurrence of each fading process; and it is observed from the BER resuts that the propagation impairment of the LMSC fading channe is reativey ower than that of Rayeigh and Loo fading channes but higher than the Rician fading channe. Keywords: mobie, mutipath fading, propagation channe, sateite communication, wideband.. Introduction The use of artificia sateites for teecommunications has received increased attention for various appications such as fixed or mobie, stratospheric patforms, navigation or broadband systems []; and the roe of sateites is changing from the traditiona teephony and teevision (TV) broadcast services to user oriented data services []. The growing demand for highy-mobie mutimedia access, such as Digita Video Broadcasting Sateite Services to Handhed devices (DVB-SH), at high data rates via the internet and other networks irrespective of the user s geographica ocation makes the roe of mobie sateite communications very significant. Mobie Terminas (MTs) are designed to communicate with the sateite whie in motion and can be cassified as and mobie, aeronautica mobie or maritime mobie, depending on their ocations on or near the earth surface [3]. The unpredictabe nature, due to time-varying radio channe, however, paces a imitation on the achievabe data rates and quaity of service (QoS) of the and mobie sateite communication (LMSC) systems. Like the terrestria inks, the LMSC inks too coud be frequency seective and time varying. Frequency-seective fading occurs if the maximum excess deay experienced by a signa 7

2 Vo 3, No., is greater than the symbo period. This resuts in rapid fuctuation in the ampitude, phase and ange of the arriving signa [4], and invariaby resuts in signa distortion. However, the LMSC inks are not exacty simiar to the terrestria ceuar inks because the eve of propagation impairments experienced by the LMSC inks depends on the geographica ocation, eevation anges and operating frequency band [5]. The need to investigate the characteristics of the LMSC channe is therefore very important and most of the existing works are based on singe- and two-state modes. The avaiabe statistica modes for LMSC channes can be categorized into two namey: singe-state modes and muti-state modes [6]. A singe-state LMSC channe mode known as the Loo mode is a singe statistica mode characterized by og-norma probabiity density function (pdf) and Rayeigh pdf and finds appication in rura environments [7, 5]. The og-norma distribution modes the shadowing attenuation affecting the ine-of-sight (LOS) signa due to foiage whie the Rayeigh distribution describes the diffuse mutipath components. The resuting compex received signa enveope is a sum of Rayeigh and og-norma processes. Corraza-Vataaro mode is another singe-state mode that combines Rician and og-norma distributions used to mode the effects of shadowing and LOS component [8]. The received signa enveope is assumed to be a product of Rician and og-norma processes; and the mode is suitabe for medium-earth orbit (MEO) and ow-earth orbit (LEO) inks [5]. [9] proposed an extension of Suzuki mode for frequency non-seective (or fat) sateite communication channes, by considering that for most of the time a LOS component is present in the received signa. The mode is a product of Rician and og-norma pdfs. The components of the Rician process are mutuay correated. In the case of non-stationary scenarios when either the sateite or MT or both move in a arge area, the propagation characteristics of such environments are appropriatey characterized by the muti-state modes [5]. The Lutz mode is a two-state (good and bad states) statistica mode based on data obtained from measurement campaigns in different parts of Europe at eevation anges between 3 o and 43 o. The good state represents LOS condition which foows Rician distribution, whie the bad state foows Rayeigh distribution []. Higher-state LMSC channe modes are based on Markov modeing approach [, 5]. A Markov process is a stochastic process in which a system takes on discrete states. [] proposed a three-state statistica channe mode after performing L-band measurements in Japan at an eevation ange of 3 o []. The first state is LOS condition, described by Rician distribution; the second state is shadowing condition, described by og-norma distribution; and the third state is the bocked condition, described by Rayeigh distribution. Most of the avaiabe LMSC channe modes are narrowband modes appicabe to second-generation (G) and third-generation (3G) systems. However, this paper focuses on the deveopment of broadband channe modes, which are more appropriate for the propagation channe characterization for future generations of LMSC systems that are expected to provide mutimedia and internet services.. Derivation of the State Occurrence Probabiity Functions 7

3 Vo 3, No., Let p A, p B and p C be the state occurrence probabiity functions for Rician, Loo and Rayeigh fading respectivey. The LMSC channe condition is a function of the assumed state occurrence probabiities as shown in Figure. Assuming the MT is in a rich scattering environment, that is with non ine-of-sight (NLOS), as in urban or suburban environments, the LMSC ink experiences mutipath Rayeigh fading for most of the time, p C is assumed to be.4, whie the remaining 6% is eft between Rician and Loo distributions to share. The probabiity of occurrence of both Rician and Loo distributions can be given as: p p =.6 () A + B The probabiity of occurrence for the Rician distribution, assuming the minimum eevation ange of the sateite is o, is computed as []: ( 9 ) 4 q =.66 impies p A. 39 p = = 64 () A q q with. Next, the probabiity of occurrence for the Loo distribution is computed by (), which gives p B =.. In order to make the mode fexibe, p A can be varied within the range of.39 and. and p within the range of. and.39. The resuting pdf of the B received signa enveope is a inear combination of the Rician, Loo and Rayeigh distributions given as: p( g or v or r) p ( g) + p ( v) p ( r) = (3) A B + where g is the Rician distribution, r is the Rayeigh distribution and v is the distribution of the Loo mode, which is the sum of og-norma and Rayeigh distributions. The Rician distribution accounts for the effect of ionospheric scintiation. The Rayeigh distribution describes the NLOS effect in the terrestria mobie channe. The shadowing effect, which is the attenuation caused by the vegetation and foiage, is described by the Loo distribution. 3. Frequency-Seective LMSC Channe Modeing The LMSC channe simuator presented in this paper is based on a three-state mode which consists of LOS or cear (state ), shadowing (state ) and deep fading (state 3); and are described by Rician, Loo and Rayeigh distributions, respectivey. A muti-state LMSC channe mode is assumed because the reative motion and ocation of the transmitter and the receiver woud resut in varying propagation behaviour. The channe is characterized by wideband effects caused by mutipath fading. The wideband effects of the LMSC channe can be we represented by the channe impuse response (CIR) h [ n, τ ] of a discrete-time mutipath fading channe mode that can be expressed as: L p h [ n, ] = h [ n ] δ [ n τ ] = L p is the number of resovabe paths; n is the discrete-time index, 73 C τ (4) where τ is the discrete reative deay of the th path; δ [ ] denotes Dirac deta function. The scattered waves with their respective deays are summed together to give the

4 Vo 3, No., impuse response of the channe. A modified Carke-based channe mode is used for the fading channe process [3]. 3. Rayeigh Fading Process The Rayeigh fading process r for the th path can be expressed as: with r jβ ( t ) = A e (5) β ω cosα + φ = (6) m and = N s i = πi + θ i N α (7) s where ω m ( = πf m ) is the maximum anguar Dopper frequency shift; A (t) is the fade ampitude (or th attenuation); α (t) and φ (t) are the ange of arriva and phase of the path, respectivey, and are statisticay independent and uniformy distributed over [-π, π); f m is the maximum Dopper frequency shift, which is given as: f vf c c m = (8) where f c is the carrier frequency, c is the speed of eectromagnetic wave and v is the speed of the MT. N sinusoids are assumed to generate the fading process, (5) can be rewritten as: s Expressing (9) in inphase and quadrature form gives: ( cos t j t ) N s r = A, i + sin β, i i β (9) r ( t ) = r ( t ) jr ( t ) () I + r (t) is the normaized ow-pass fading process whose pdf is Rayeigh. The fade enveope of (t) Q r is then obtained as: and the phase is: { ri rq } r = R () + r ( ) Q t φ ( ) = tan t () ri where R{ } denotes the rea part. (9) is a modified Carke s mode which is a Wide-Sense Stationary Uncorreated Scattering (WSSUS) Rayeigh fading simuation mode. The fading process is generated for a the paths and the 74

5 Vo 3, No., resutant fading process is the sum of a the L p fading processes. 3. Rician Fading Process Simiary, the normaized ow-pass fading simuation mode for the Rician fading can be expressed as: with g ( t ) = g ( t ) jg ( t ) (3) I + { ri + K cos( mt cosθ + φ )} K { r + K sin( t cosθ + φ )} K gi = + Q ω (4) ω (5) gq = Q m + where K is the Rice factor, which represents the magnitude of the LOS signa; θ and φ are the ange of arriva and the initia phase, respectivey, of the LOS component. 3.3 Loo Fading Process The Loo simuation mode is an addition of ognorma and Rayeigh processes which can be expressed as: v I and {[ ( v + r ) + j( v + r )] exp( jπf )} v R t = (6) I I Q v Q are ognorma inphase and quadrature components, respectivey. The moderate shadowing is assumed and the vaues of the mean (µ) and standard deviation (σ) for ognorma distribution used in the simuation are -.5 and.6, respectivey, given by [4]. Q c 3.4 Simuation Method A tapped-deay ine of six paths (L p = 6), typica of wideband fading effect of mobie communications, is used to mode the frequency-seective fading channes. Each tap is described by its time-varying gain a and the corresponding deayτ. The deay set for the six paths is τ =,,,3,4, 5 sampes. The reative ampitudes (or attenuations) of the mutipath components are assumed to foow the exponentia power deay profie expressed as: a 6 = exp( τ 3) exp( τ 3), (7) i= i Using the state occurrence probabiities, a set of states consisting of s, s and 3s is randomy generated and sorted in ascending order. Then for a particuar run, the simuator generates the fading process corresponding to the state number as the LMSC channe, which coud be Rician, Loo or Rayeigh fading channe. Separate fading signa is generated for each of the 6 paths in order to make a path independent from another. The Agorithm describes the 75

6 Vo 3, No., process of generating the frequency-seective LMSC fading channe. The fading is appied to the transmitted signa through convoution. Agorithm : LMSC fading channe END. Inputs: path deays and attenuations sets. Outputs: fading channe 3. BEGIN 4. Get current state number 5. IF state is, generate Rician fading; END IF 6. IF state is, generate Loo fading; END IF 7. IF state is 3, generate Rayeigh fading; END IF 8. FOR k to L p 9. Generate the fading process corresponding to the state number. Appy attenuation[k] to fading signa. Appy deay[k] to fading signa. Sum fading signa for each k path Simuation Parameter Specification Signaing scheme OFDM-QPSK Signa-to-noise ratio to 8 db Number of paths 6 Fading spectrum Modified Carke s mode Number of fading sinusoids 6 Number of channe reaizations Path deays (sampes) [,,, 3, 4, 5] Average path gains (db) [ ,-6.9, ,-9.864,-.634,-.87] Rice factor (K) 3 Ange of arriva of LOS component π/4 Ange of eevation of sateite o Mobie speeds km/h, 5 km/h Carrier frequency. GHz 4. Resuts and Discussion The simuation of the fading channes was carried out in MATLAB software environment. MATLAB codes 76

7 Vo 3, No., were written for the impementation of the modes and run to generate resuts in form of graphs for visuaization. The comparison between the LMSC fading channe and each of Rician, Loo and Rayeigh fading channes are done in terms of bit error rate (BER) performance of an OFDM-QPSK signaing scheme for signa-to-noise ratio (Eb/No) of to 8 db at mobie speed of km/h and state occurrence probabiities of p A =.39, p B =. and p C =.4 for Rician, Loo and Rayeigh fading respectivey. Tabe shows the state occurrence probabiities for different propagation conditions. This means that at any time instant, the LMSC ink can experience a transition from one state to another. The ower the probabiity of occurrence of any distribution, the esser the contribution of the distribution to the resutant LMSC fading channe. The exponentia power deay profie of the 6-path frequency-seective LMSC channe is shown in Figure. The average received power (in db) reduces as the period of the deay (in µs) increases. This impies that the deay experienced by any specific path resuts in oss of the signa power. Figures 3(a) and 3(b) show the power spectra density for mobie speed of 5 km/h and km/h respectivey. The resuts show that the higher the Dopper frequency the broader the fading spectrum. That is, as the speed of the mobie termina increases, there is a corresponding increase in the maximum Dopper frequency shift which invariaby widens the fading spectrum. Figures 4(a) and 4(b) show the frequency response of the LMSC channe for mobie speed of 5 km/h and km/h respectivey. The higher spikes observed for the speed of 5 km/h as compared to km/h reveas that the higher the mobie speed, the higher the variation in the frequency response of the channe with time as a resut of increase in the Dopper frequency shift. Figures 5 and 6 show the pdf of the Rayeigh and Rician fading process respectivey. The effect of varying the Rice factor K which is the specuar (or LOS) component is shown. It is observed from the graph that if K =, the Rician pdf reduces to the Rayeigh pdf, which impies that there is NLOS between the transmitter and the receiver. The pdf of the Loo (Rayeigh + og-norma) fading process is shown in Figure 7. Figure 8(a) shows the pdf of the LMSC channe fading distribution, which is a combination of the Rayeigh, Rician and Loo; with the probabiity of occurrence of.39,. and.4 for the Rician, Loo and Rayeigh respectivey. The contribution of the Rayeigh pdf to the resutant pdf of the LMSC channe is the highest, foowed by the Rician pdf and the Loo pdf gives the owest contribution. Conversey, the probabiity of occurrence of.,.39 and.4 for the Rician, Loo and Rayeigh respectivey give the LMSC channe fading distribution pdf shown in Figure 8(b). Here, the Loo pdf gives higher contribution than the Rician pdf. The reason why the Rayeigh fading contribution is made the highest is based on the assumption that the LMSC ink experiences NLOS most of the time. This is true for typica urban and suburban environments. Figures 9 and show the simuated LMSC channe fading at 5 km/h and km/h, respectivey. The resuts show that the fading signa eve varies more rapidy at 5 km/h but sower at km/h. In Figure, the Rician channe gives a mean BER of.69 whie the LMSC channe gives a mean BER of.85. The AWGN channe, the idea channe condition needed for reiabe LMSC ink, gives a mean BER of.8. The resut from Figure reveas that the Loo channe gives a mean BER of.4 as against.85 for the LMSC channe. Aso, from Figure 3, the Rayeigh channe gives a mean BER of. compared to.85 for the LMSC channe. The comparison of a the fading channes is presented in Figure 4. The mean BER resuts show that the Rayeigh channe produces the worst 77

8 Vo 3, No., propagation impairments foowed by the Loo channe, and then foowed by the LMSC channe, whie Rician channe gives the east propagation impairments. It is observed in Figure 5 that for p A =.39, p B =. and p C =.4, the mean BER is.85; and the LMSC ink is considered cear as it experiences Rician fading more of the time than the Loo fading. This is because for the Rician channe, there is a strong LOS path among the signa paths; whie for the Loo channe, a the paths experience shadowing. For p A =., p B =.39 and p C =.4, the mean BER is.93 and the LMSC ink is said to experience more shadowing effect than LOS. When p A =., p B =.3 and p C =.6, the mean BER is.78. This is a deep fading scenario where the LMSC ink experiences mutipath Rayeigh fading (with NLOS) amost the entire time share, typica of an environment with many bockages in the signa path. 5. Concusion and Recommendations 5. Concusion The deveopment of a simuation mode for frequency-seective and mobie sateite communication channe has been anayzed, and the performance characteristics of the channe mode have been evauated in terms of bit error rate. The mode assumes a three possibe channe states namey cear, shadowing and deep fading. The resuts obtained from computer simuation show that the degree of propagation impairments of the LMSC channe depends on whether the channe is experiencing Rician fading (cear), Loo fading (shadowing) or Rayeigh fading (deep fading). The proposed simuation mode is fexibe because by varying the state occurrence probabiities: p A (for Rician), p B (for Loo) and p C (for Rayeigh), the mode is appicabe to the moderate suburban, suburban and urban environments. Thus, the proposed LMSC channe mode presented in this paper demonstrates the possibiity of evauating the performance of a and mobie sateite communication system based on the propagation environment under consideration. 5. Recommendations For further works, the proposed mode is open to modifications and empirica vaidations because it coud be considered as a heuristic approach. This work woud serve as a usefu reference for researchers and engineers in the fied of mobie and sateite communications. References [] Castanet, L. (). Channe modeing and propagation impairment simuation activities within the SatNEx project, Internationa Journa of Sateite Communications and Networking, vo. 9, no.,, -6. [] Mir, R. M. (), Sateite Data Networks, Technica report. [Onine] Avaiabe: (Juy 6, ) [3] Icev, S. D. (5). Goba Mobie Sateite Communications: For Maritime, Land and Aeronautica Appications, Netherands: Springer Ltd. (Chapter 5). [4] Skar, B. (997). Rayeigh fading channes in mobie digita communication systems, IEEE Communications Magazine, 9-. [5] Asad, M. & Abbas, M. (), Characterization and channe modeing for sateite communication systems, Sateite Communications, Bekinge Institute of Technoogy, Sweden, [Onine] Avaiabe: 78

9 Vo 3, No., (Juy 6, ) [6] Abdi, A., Lau, C. W., Aouini, M. & Kave, M. (3). A new simpe mode for and mobie sateite channes: first- and second-order statistics. IEEE Trans. Wireess Comm., vo., no. 3, [7] Loo, C. & Secord, N. (99). Computer Modes for Fading Channes with Appications to Digita Transmission, IEEE Transactions on Vehicuar Technoogy, vo. 4, no. 4, [8] Corraza, G. E. & Vataaro, F. (994). A statistica channe mode for and mobie sateite channes and its appication to nongeostationary orbit systems, IEEE Trans. Vehicuar Technoogy, vo. 43, no. 3, [9] Patzod, M., Kiat, U. & Laue, F. (998). An extended Suzuki mode for and mobie sateite channes and its statistica properties, IEEE Trans. Vehicuar Technoogy, vo. 47, no, [] Lutz, E. Cygan, D. Dippod, M. Donasky, F. & Papke, W. (99). The and mobie sateite communication channe - recording, statistics and channe mode, IEEE Transactions on Vehicuar Technoogy, vo. 4, no, [] Karaiopouos, M. S. & Pavidou, F. N. (999). Modeing the and mobie sateite channe: a review, Eectronics and Communication Engineering Journa, [] Karasawa, Y., Kimura, K. & Minamisono, K. (997). Anaysis of avaiabiity improvement in LMSS by means of sateite diversity based on three-state propagation channe mode, IEEE Trans. Vehicuar Technoogy, vo. 46, no. 4, [3] Xiao, C. & Zheng, Y. R. (3). A Statistica Simuation Mode for Mobie Radio Fading Channes, IEEE Transactions, [4] Loo, C. & Secord, N. (99), Computer Modes for Fading Channes with Appications to Digita Transmission, IEEE Transactions on Vehicuar Technoogy, vo. 4, no. 4, pp Dr. Z.K. Adeyemo received the B.Eng. and M.Eng. degrees in Eectrica Engineering from University of Iorin, Iorin, Nigeria and his Ph.D. degree in Eectronic and Eectrica Engineering from Ladoke Akintoa University of Technoogy (LAUTECH), Ogbomoso, Nigeria. He is a member of the IEEE and a registered member of Counci for the Reguation of Engineering in Nigeria (COREN). His research interest is on signa processing in mobie communications. Oumide O. Ajayi (M 7) became a student member of the IEEE in 7. He received the B.Tech. degree in Eectronic and Eectrica Engineering from LAUTECH, Ogbomoso, Nigeria in 8. He is a teaching assistant and currenty pursuing his M.Tech. degree in the same university. His research interests incude wireess communications, information security and adaptive signa processing. Festus K. Ojo received the B.Tech. degree in Eectronic and Eectrica Engineering from LAUTECH, Ogbomoso, in 8 and his M.Eng degree in Communication Engineering from Federa University of Technoogy (FUTA), Akure, Nigeria in. His research interest is on digita communications. 79

10 Vo 3, No., Tabe. State occurrence probabiity of Rice, Loo and Rayeigh Distributions Rician Loo Rayeigh LMSC p A p B p C channe state cear shadowing..3.6 deep fading -4 Mutipath Power Deay Profie -5 Average Received Power (db) Path Deay (us).8. x -5 Figure. LMSC channe mode as a function of state occurrence probabiity Figure. Power Deay Profie of the Frequency- Seective LMSC Fading 8

11 Vo 3, No., x 6 PSD of the LMSC channe at mobie speed of 5 km/h x 6 PSD of the LMSC channe at mobie speed of km/h Power Spectra Density Power Spectra Density Frequency x 5 Figure 3(a). Received spectrum for the Carke-based LMSC channe mode at 5 km/h Frequency x 5 Figure 3(b). Received spectrum for the Carke-based LMSC channe mode at km/h F re q u e n c y R e s p o n s e (d B ) Frequency Time F req u e n c y R e s po n s e (d B ) Frequency Time Figure 4(a). Frequency Response of the LMSC fading channe for 5 km/h Figure 4(b). Frequency Response of the LMSC fading channe for km/h 8

12 Vo 3, No.,.9 Simuation Theoretica.8 K=.8.6 Probabiity Density Function (pdf) Probabiity Density Function K=5 K=3 K= K= (Rayeigh) Fade Enveope (r) Figure 5. PDF for the Rayeigh fading distribution 3.5 Loo Fading Distribution Fade Enveope (g) Figure 6. PDF for the Rician fading distribution 3 Probabiity Density Function (pdf).5.5 Probabiity Density Function (pdf) Fade Enveope (v) Fade Enveope (h) Figure 7. PDF for the Loo fading distribution Figure 8(a). PDF for the LMSC channe fading distribution with [p A =.39, p B =., p C =.4] 8

13 Vo 3, No., A B C A B C Probabiity Density Function (pdf) F a d in g S ig n a L e v e (d B ) Mean =-9. db Fade Enveope (h) Eapsed Time (us) Figure 8(b). PDF for the LMSC channe fading distribution with [p A =., p B =.39, p C =.4] Figure 9. Simuated LMSC channe fading at 5km A B C OFDM-QPSK Signa Transmission over Mutipath Fading Channes AWGN channe Rician channe LMSC channe F a d i n g S ig n a L e v e ( d B ) B E R Mean = db Eapsed Time (us) Figure. Simuated LMSC channe at km/h EbNo Figure. BER comparison of the Rician and LMSC channes at km/h 83

14 Vo 3, No., OFDM-QPSK Signa Transmission over Mutipath Fading Channes AWGN channe Loo channe LMSC channe OFDM-QPSK Signa Transmission over Mutipath Fading Channes AWGN channe Rayeigh channe LMSC channe - - B E R B E R EbNo Figure. BER comparison of the Loo and LMSC channes at km/h EbNo Figure 3. BER comparison of the Rayeigh and LMSC channes at km/h - AWGN Rician Rayeigh Loo LMSC - LMSC [.39,.,.4] LMSC [.,.39,.4] LMSC [.,.3,.6] Deep fading B E R B E R Shadowing - - Cear EbNo Figure 4. BER comparison of a the fading Channes at km/h EbNo Figure 5. Effect of the state occurrence probabiities on the LMSC channe at km/h 84