Journal of Engneerng Scence and Technology Vol., o. 4 (6) 476-495 School of Engneerng, Taylor s Unversty A IMPROVED BIT LOADIG TECHIQUE FOR EHACED EERGY EFFICIECY I EXT GEERATIO VOICE/VIDEO APPLICATIOS VIOTH BABU K.*, RAMACHADRA REDDY G., ARTHI M. School of Electroncs Engneerng, VIT Unversty, Vellore Campus, Vellore, 63 4, Taml adu, Inda *Correspondng Author: vnothbabu.k@vt.ac.n Abstract Mult nput mult output (MIMO) and orthogonal frequency dvson multplexng (OFDM) are the key technques for the future wreless communcaton systems. Prevous research n the above areas manly concentrated on spectral effcency mprovement and very lmted work has been done n terms of energy effcent transmsson. In addton to spectral effcency mprovement, energy effcency mprovement has become an mportant research because of the slow progressng nature of the battery technology. Snce most of the user equpments (UE) rely on battery, the energy requred to transmt the target bts should be mnmzed to avod quck battery dran. The frequency selectve fadng nature of the wreless channel reduces the spectral and energy effcency of OFDM based systems. Dynamc bt loadng (DBL) s one of the sutable soluton to mprove the spectral and energy effcency of OFDM system n frequency selectve fadng envronment. Smple dynamc bt loadng (SDBL) algorthm s dentfed to offer better energy effcency wth less system complexty. It s well suted for fxed data rate voce/vdeo applcatons. When the number of target bts are very much larger than the avalable subcarrers, the conventonal sngle nput sngle output (SISO)-SDBL scheme offers hgh bt error rate (BER) and needs large transmt energy. To mprove bt error performance we combne space frequency block codes (SFBC) wth SDBL, where the adaptatons are done n both frequency and spatal doman. To mprove the qualty of servce (QoS) further, optmal transmt antenna selecton (OTAS) scheme s also combned wth SFBC-SDBL scheme. The smulaton results prove that the proposed schemes offer better QoS when compared to the conventonal SISO- SDBL scheme. Keywords: Energy effcency, OFDM, Optmal transmt antenna selecton, Smple dynamc bt loadng, Space frequency block codes. 476
An Improved Bt Loadng Technques for Enhanced Energy Effcency n.... 477 omenclatures B T b(m) b max D E T e(m) H(m) L T P S,S Target total bts to be transmtted through one OFDM symbol umber of bts on m th subcarrer Maxmum number of bts that can be loaded on a subcarrer SFBC matrx Total energy requred to transmt the target bts Energy requred for m th subcarrer Channel gan on m th subcarrer Total number of subcarrers Total number of transmttng antennas umber of groups Selected antenna ndces Abbrevatons CFO ISI LTE LTE-A AS W-MAX. Introducton Carrer Frequency Offset Inter Symbol Interference Long Term Evoluton Long Term Evoluton- Advanced o Antenna Selecton Worldwde Interoperablty for Mcrowave Access Broadband data connecton through UE s the order of the day. The usage of Personal Computers (PC) for nternet access s reducng and the usage of moble phones for nternet access s growng day by day []. In countres lke Chna, Korea, Unted States of Amerca (USA) and Japan 8% of nternet access s through moble phones. 5% of the moble users use moble as the prmary nternet source. Hgh speed meda streamng applcatons lke YouTube, Facebook, Google+ attracted huge amount of customers. 65% of tme spent on socal networks s through moble phones. 6 bllon hours of vdeo are watched through YouTube every month. More than 3% of YouTube s global watch tme s through moble phones. Recent days moble bankng has become very popular. In 3 globally, 59 mllon moble users used ther moble for bankng purposes. The number s expected to cross bllon durng 7. In 3, 8 of top fnancal nsttutons n USA offered moble bankng servces. Skype s very popular for vdeo calls through nternet. Around 33 mllon hours people talkng, laughng, jokng, sngng through Skype n a day. Almost 65% of Skype usage s through mobles. People also use moble phones as Wreless Fdelty (W-F) hot spot. All the above examples and the related statstcs clarfy the necessty of moble broadband. The customers are expectng hgh qualty servces from moble operators. Mantanng acceptable BER or SR at the recever s synonymous to QoS at the physcal layer level []. To acheve the above demands transmsson parameters lke modulaton order and energy are dynamcally adopted as per the channel qualty. Mantanng wthn maxmum delay and mnmum acceptable rate s synonymous to QoS at the hgher layer levels. QoS at hgher layers are possble by proper channel allocaton and schedulng methods. Energy
478 K. V. Babu et al. consumpton s an mportant metrc for QoS mprovement []. In ths work, QoS mprovement by proper energy allocaton at physcal layer level s consdered. Zero ISI and one tap equalzaton features attracted the researchers to use OFDM for 4G standards lke W-MAX, LTE and LTE-A. Usage of OFDM for hgh qualty multmeda servces under fadng envronment s a challengng task. Rado channels are passve crcuts where large amount of energy s wasted [3]. Popularty of smart mobles and ts servces lke vdeo games, vdeo conferencng, Hgh Defnton Televson (HDTV), web semnars, e-mal and downloadng fles consumes energy at a hgher rate than the normal mobles. Hgh performance recevers are expected to offer hgh spectral effcency []. Ths ncreases the system complexty and t also needs costler components. Hgh performance recevers consume more energy. Increase n multmeda applcatons ncrease the sgnal processng complexty further. Snce most of the moble devces are operatng wth the battery, hgh energy consumpton wll cause the battery to dry soon. Advanced physcal layerng technques lke ICI mtgaton, DBL,MIMO, dynamc guard nterval, blnd channel and CFO estmaton are requred to ncrease the QoS of OFDM system wthout ncreasng the system complexty []. The deeply faded subcarrers contrbute to more BER, ncreasng average BER of OFDM symbol beyond the target BER [4]. To mantan the same BER, the deeply faded subcarrers have to be dentfed and omtted from transmsson. Ths mproves BER performance wth slght loss n spectral effcency. Ths spectral effcency loss can be compensated by effectvely employng DBL. The frequency selectve fadng nature of the wreless channel reduces the spectral and energy effcency of OFDM based systems. DBL s the powerful scheme to mprove the spectral and energy effcency of OFDM system n frequency selectve fadng envronment [5]. To mantan QoS n energy allocaton based algorthms, more energy s allocated for poor subcarrers and less energy for good subcarrers. Ths makes the chances of successful recepton hgh. Thus DBL algorthms mprove the QoS. The delay senstve transmsson servces lke voce and vdeo are usually served wth fxed rate. Margn Adaptve (MA) algorthms are better suted for fxed rate applcatons. These algorthms try to maxmze margn or mnmze energy requred to transmt the target bts wth the target of fxed data rate. It can be mathematcally represented as [6], Mnmze E m so that L m= L T = e( ) m= b( m) = B and b( m) b T max () where B T s the target total bts to be transmtted through one OFDM symbol and E T s the total energy requred to transmt the same. b max s the maxmum number of bts that can be loaded on a subcarrer. e(m) and b(m) are respectvely the number of bts per subcarrer and energy requred to transmt the same. There are many MA algorthms dscussed n the lterature.
An Improved Bt Loadng Technques for Enhanced Energy Effcency n.... 479 For fast fadng condtons, coherence tme s small [7] compared to the total delay nvolved for channel estmaton, feedback and DBL algorthm convergence. Snce the channel becomes outdated, the parameters adopted based on ths wll go wrong. Ths reduces the QoS of the system [4]. To offer superor qualty servces for long duratons an optmal DBL algorthm whch takes less teraton for converge s the major requrement []. Esfahan and Afrasab [6] proved that SDBL algorthm offers better performance than the conventonal optmal and sub optmal algorthms lke Hughes-Hartogs [8], Fscher and Huber [9] and Chow []. SDBL algorthm optmally allocates bts to all subcarrers n teraton whereas the other above mentoned algorthms need more teraton to converge. It s proved that ntal bt loadng result of SDBL s nearer to Hughes-Hartogs fnal bt loadng result. But Hughes-Hartogs algorthm converges after teratons. The algorthm works well when the number of subcarrers s suffcent (Case ) and more (Case ) when compared to the target bts. An mportant ssue s when the target bts are very large compared wth the number of avalable subcarrers (Case 3); the conventonal SISO-SDBL scheme offers hgh BER. It also consumes more energy to transmt the target bts. In ths case, SDBL algorthm forcefully allocates the target bts to lmted number of subcarrers rrespectve of the channel frequency response. Ths ncreases the probablty of error. To acheve a certan target BER, t requres more transmt energy. Usually the thrd case wll not converge n an teraton for poor channel condtons. Snce overall am of ths work s to reduce the total transmt energy and to mprove QoS, an optmum soluton s requred to effcently use SDBL for fxed rate applcatons. A combnaton of SFBC and antenna selecton schemes wth SDBL s used to solve the above ssue. We modfy the algorthm to sut for SFBC based MIMO-OFDM system. In Mult Carrer Modulaton (MCM) scheme such as OFDM, DBL s carred out n frequency dmenson. Recent research on DBL n OFDM has proved that the hybrd adaptaton based on frequency and space dmenson s expected to offer better BER performance than the conventonal adaptaton whch s done only on frequency dmenson [5]. MIMO combned wth OFDM offers dversty gan whch reduces the BER of the system []. A sutable dversty scheme coupled wth the conventonal SDBL algorthm wll solve the problems of hgh BER and large transmt energy. Receve dversty schemes such as Maxmal Rato Combnng (MRC) offers better dversty gan. But lmtatons to deploy multple antennas at UE make MRC neffcent. Transmt dversty technques based on beam formng offers performance smlar to MRC. But ths scheme requres perfect knowledge about the channel at the transmtter sde [5]. Orthogonal Transmt Dversty (OTD) schemes such as Space Tme Block Codes (STBC) perform almost smlar to transmt dversty based on beam formng [5, ]. These schemes do not requre perfect channel knowledge at the transmtter sde. But STBC s more suted for flat fadng channels.we prefer SFBC based dversty scheme whch s well suted for frequency selectve fadng channels [3]. To mplement SFBC scheme, mnmum two transmttng antennas are requred. SFBC mplemented wth more transmttng antennas wll ncrease the dversty gan at the cost of reduced spectral effcency and ncreased complexty.
48 K. V. Babu et al. Other lmtatons lke number of Rado Frequency (RF) chans and synchronzaton make SFBC wth more transmttng antennas mpractcal. Instead of usng more transmttng antennas, an alternate approach OTAS scheme s used n ths work to mprove the BER performance of SFBC scheme [4-6]. Ths scheme selects two antennas havng good channel frequency response out of 4 or 8 transmttng antennas for every coherence perod. Bt allocaton s done based on the selected antennas. Snce the proposed system always use two best transmttng antennas, the energy requred to transmt the target bts gets reduced. The rest of the paper s organsed as follows. In Secton the proposed system model s brefly explaned. Secton 3 gves bref descrpton on OTAS. Secton 4 provdes explanaton on the proposed MA-SFBC-SDBL-OTAS algorthm. In Secton 5, smulaton results are dscussed and Secton 6 concludes the paper.. System Model The proposed system uses a hybrd confguraton of SFBC, OTAS and SDBL. Whle OTAS provde wth better channels, SFBC provde spatal and frequency dversty gan to the system. The SDBL performance s mproved dramatcally wth ths hybrd confguraton. The user data n form of bts s modulated by the M-ary Quadrature Ampltude Modulaton (M-QAM). The modulated symbols for the I th OFDM symbol of length L can be gven as [ ] X [ m] = X [], X [], X []... X [L ] T () I I I I I L where m=,,...,, T represents transpose operaton. These symbols are nput to the SFBC block. Here we use SFBC for X system. In SFBC, Alamout scheme s mplemented over neghbourng subcarrers of the same OFDM symbol. The SFBC coded data s gven as [, 7] X I [ m] X I [m+ ] D= = [ D * * D ] X I [m+ ] X I [ m] The symbols coded n the frst (D ) and second (D ) columns of the matrx D are transmtted through the selected two transmt antennas respectvely. It s clear that each modulated symbol s transmtted n two dfferent frequences and antennas. It adds spatal and frequency dverstes to each modulated symbol. In SFBC-OFDM system, for each subcarrer both antennas sub channel gans are consdered. For each receved subcarrer at least one antenna s subchannel gan may be good whch may lead the system to use hgher order modulaton for that subcarrer. Same number of bts s allocated for each subcarrer n both the transmttng antennas to reduce the decodng complexty. The major buldng blocks of SFBC- SDBL-OTAS scheme s shown n Fg.. The channels are estmated at the recever and the antenna selecton s carred out. Based on the selected antennas channel gans, the number of bts that can be loaded on each subcarrer (modulaton ndex, M) s dentfed usng SDBL algorthm. The OTAS ndex and modulaton ndex s fed back to the transmtter (3)
An Improved Bt Loadng Technques for Enhanced Energy Effcency n.... 48 usng the feedback channel [9]. The nput bts are modulated usng the receved modulaton ndex and the two SFBC modulated OFDM symbols are transmtted through the selected antennas. The SFBC-SDBL scheme whch selects out of 4 and 8 transmttng antennas s named as SFBC-SDBL-OTAS (4) and SFBC- SDBL-OTAS (8). Fg.. Block dagram of the proposed MA-SFBC-SDBL-OTAS bt loadng scheme. 3. OTAS When the number of transmttng antennas ncreases, the complexty also ncreases along wth the dversty gan. The system becomes expensve snce each antenna needs separate RF chans [3]. However when MISO system s used n dversty mode, more number of transmttng antennas wll reduce the spectral effcency. SFBC wth two transmttng antennas lead to Rate system, where two symbols are transmtted n two subcarrers. Whereas SFBC wth three transmttng antennas leads to Rate.5 system, where four symbols are transmtted n eght subcarrers, thereby reduces the spectral effcency. For SFBC wth three transmttng antennas, eght consecutve subcarrers should have same channel gan. Ths quas statc condton s volated under fast fadng [4, 5]. Also, accurate synchronzaton of multple transmttng antennas wth one recevng antenna ncreases the complexty. Moreover, feedback sgnalng used for the synchronzaton, causes wastage of bandwdth. Therefore, nstead of usng all the transmttng antennas use two antennas wth best channel frequency response for every coherence perod. It has been proved that antenna selecton schemes offers better spectral effcency than the conventonal system wth no antenna selecton [4]. OTAS s expected to reduce the hardware and sgnal processng complexty of conventonal MISO systems wth no antenna selecton. The followng steps are executed n the OTAS scheme.. Calculate Frobenus norm for the channel frequency response avalable at each transmtter antenna [5].
48 K. V. Babu et al. L,S,S I = I ( ), =,... T m= H H m for S (4) where T represents the total number of transmt antennas avalable.. Two antennas out of T transmtter antennas whch have hghest Frobenus norms are then selected as [4]., S, S { S, S} arg max { H I H I } = + (5) S, S T S S In the conventonal SFBC-SDBL system, f any one of the sub channel s bad, the average channel gan becomes less and forces the system to use lower order modulaton whch agan reduces the QoS. To mplement SFBC-SDBL always wth good channel condtons OTAS scheme s combned wth SFBC-SDBL scheme. 4. MA-SFBC-SDBL-OTAS Algorthm In SDBL algorthm, the followng steps are executed n sequence to dentfy the optmum number of bts per subcarrer.. Fnd sub channel gan H as, S, S H I ( m) + H I ( m) H ( m) = ; m=,l L (6). Partton the sub carrers nto P groups usng max H P= log + mn H (7) where mn H and max H are mnmum and maxmum sub channel gans.. The lower and upper ndex of each group s calculated usng, max H LI = P + max H UI =, =,... P P The subcarrer whch s havng the channel gan twce the other sub carrer can carry one more bt than the other subcarrer for the same energy and SR gap. From (8), t s clear that upper ndex of each group s twce the lower ndex. Suppose the subcarrers n the frst group are loaded wth one bt, then the subcarrers n the second group can be loaded wth two bts. For every group one more addtonal bt can be allocated. v. Identfy the number of subcarrers fall n each group. The number of subcarrers fall n th group ( ) s calculated by countng the sub carrers havng channel gans between LI and UI. Same procedure s repeated to dstrbute L subcarrers to all P groups. (8)
An Improved Bt Loadng Technques for Enhanced Energy Effcency n.... 483 P = L (9) = v. The target bts are dvded and dstrbuted based on the number of subcarrers and target bts. Ths creates three dfferent cases [6]. Case and Case are not dscussed n ths paper. We concentrate on mprovement of system performance n Case 3 where SDBL performance degrades Case (): umber of subcarrers s suffcent enough to allocate the target bts. Case (): The number of subcarrers s suffcently large when compared to the target bts.e BT <. L. P = Case (3): B T s large.e B >. + L(b P) T P = In ths case, the number of subcarrers are not suffcent for the gven target bts. ormal allocaton as per Case makes B > b whch leads to more than b max bts per subcarrer n the last group. To avod ths, allocate last group wth B = b and dstrbute the remanng target bts to other groups. P max P round[( β+ ) ] ; =,...P P B = BT bmax P B k ;= P () k= bmax P ;= P where P B b. β= L T max P = P max P max. p () v. After dentfyng the number of bts for each group, dstrbute them among the subcarrers avalable n each group usng, u B b = + l u b = b u B = B l u = In th group, b l bts are allocated to l subcarrers and b u bts are allocated to u subcarrers such that, u u l l () B = b + b (3) and + = (4) l u
484 K. V. Babu et al. 5. Smulaton Results and Dscusson Esfahan and Afrasab [6] proved that SDBL algorthm offers better performance than the conventonal optmal and sub optmal algorthms. So n ths work, we have not compared the performance of SDBL algorthm wth other algorthms. In ths work SISO-SDBL algorthm performance s set as the benchmark and we tred to mprove ts performance further by addng SFBC and OTAS schemes. The performance of the proposed algorthm s tested for all three cases of total target bts. The followng parameters are taken for the smulaton to test the performance of the proposed algorthm. System bandwdth and subcarrer bandwdth s assumed to be 5 MHz and 39.63 khz respectvely. The total number of subcarrers s 8. M-ary QAM scheme s used to modulate each subcarrer. Based on the channel condton, M value s vared from to 56. Maxmum number of bts that can be loaded on a subcarrer s lmted to 8 (b max =8). The smulaton results are repeated for dfferent channel condtons and the average energy values are dsplayed. To analyze the performance of MA algorthms, energy effcency (b/j) parameter s frequently used n the lterature [7, 8]. It s the rato between the target bts for one OFDM symbol and the total energy requred to transmt the target bts. Snce the SISO-SDBL algorthm perform good for Case and Case type of bt allocaton, the results for Case 3 are gven n ths paper. The channel n Fg. s taken to test the performance of SISO-SDBL algorthm. To test SDBL for Case 3, we take target bts to be 83. The upper and lower ndex of each group, total number of bts and subcarrers allocated to each group, number of bts allocated to the ndvdual subcarrers n each group for Case 3 s shown n Table. The bt allocaton for Case 3 s shown n Fg. 3. All 83 bts are effectvely allocated to all subcarrers. Energy allocated per subcarrer for Case 3 s shown n Fg. 4. For Case 3 of target bts, the conventonal SISO-SDBL algorthm offers an average energy effcency of 4.35 b/j for the gven target bts. For Case 3, as n Table, subcarrers n 7 th and 8 th groups are loaded wth 9 bts exceedng the maxmum bt lmt at the ntal teraton. To avod ths, the algorthm has to be reterated. Table. umber of bts assgned to the number of subcarrers n each group for SISO-SDBL scheme: Case 3. Group Lower Upper l u l u B Index () Index Index b b....444 5 3 4 3.444.887 3 4 5 4.887.774 6 3 4 4 5 5.774.3548 59 4 5 3 4 6.3548.797 5 6 8 5 3 7.797.493 37 8 9 5 3 8 8.493.8387 384 8 9 48 48
An Improved Bt Loadng Technques for Enhanced Energy Effcency n.... 485 Subchannel gan - - 4 6 8 4 Subcarrer Index Fg.. Sub channel gan vs. Subcarrer ndex. 9 8 7 Bts per subcarrer 6 5 4 3 4 6 8 4 Subcarrer ndex Fg. 3. Bts per subcarrer vs. Subcarrer ndex for SISO-SDBL scheme: Case 3. 6 5 Energy per subcarrer 4 3 4 6 8 4 Subcarrer Index
486 K. V. Babu et al. Fg. 4. Energy allocated per subcarrer vs. Subcarrer ndex for SISO-SDBL scheme: Case 3. The bt and energy allocaton for SFBC-SDBL-AS scheme s shown n Fgs. 5 and 6 respectvely. The sub channel gans are dvded nto 4 groups. The SDBL parameters obtaned for SFBC-SDBL-AS scheme s shown n Table. SFBC scheme ntroduces frequency and spatal dversty gans to the conventonal SISO-SDBL system whch mproves the QoS. The SFBC-SDBL-AS scheme offers an average energy effcency of 7.97 b/j for the gven target bts. To mprove the performance further, OTAS scheme s coupled wth SFBC-SDBL scheme. When two antennas are selected out of 4 transmttng antennas, the average energy effcency becomes 9.78 b/j for the gven target bts. The bt and energy allocaton for the SFBC-SDBL-OTAS (4) scheme are shown n Fgs. 7 and 8. Here the sub channel gans are dvded nto 3 groups. The SDBL parameters obtaned for SFBC-SDBL-OTAS (4) scheme s shown n Table 3. 8 7 6 Bts per subcarrer 5 4 3 4 6 8 4 Subcarrer ndex.8 Fg. 5. Bts per subcarrer vs. subcarrer ndex for SFBC-SDBL-AS scheme: Case 3..7 Energy per subcarrer.6.5.4.3.. 4 6 8 4 Subcarrer Index
An Improved Bt Loadng Technques for Enhanced Energy Effcency n.... 487 Fg. 6. Energy allocated per subcarrer vs. subcarrer ndex for SFBC-SDBL-AS scheme: Case 3. Table. umber of bts assgned to the number of subcarrers n each group for SFBC-SDBL-AS scheme: Case 3. Group Lower Upper l u l u B Index () Index Index b b.435.8469 44 4 5 6 4.8469.6938 35 5 6 5 5 3.6938 3.3876 4 6 7 4 35 4 3.3876 6.775 49 7 8 35 3 58 8 7 6 Bts per subcarrer 5 4 3 4 6 8 4 Subcarrer ndex Fg. 7. Bts per subcarrer vs. subcarrer ndex for SFBC-SDBL-OTAS (4) scheme: Case 3..7.6 Energy per subcarrer.5.4.3.. 4 6 8 4 Subcarrer Index
488 K. V. Babu et al. Fg. 8. Energy allocated per subcarrer vs. subcarrer ndex for SFBC-SDBL-OTAS (4) scheme: Case 3. Table 3. umber of bts assgned to the number of subcarrers n each group for SFBC-SDBL-OTAS (4) scheme: Case 3. Group Index () Lower Index Upper Index B l b u b l u.746.49 46 5 6 8 9.49 4.98 347 6 7 5 5 57 3 4.98 8.5964 439 7 8 57 5 6 When two antennas are selected out of 8 transmttng antennas, the average energy effcency becomes 65. b/j for the gven target bts. The bt and energy allocaton for the SFBC-SDBL-OTAS (8) scheme are shown n Fgs. 9 and. The sub channel gans are dvded nto 3 groups. The SDBL parameters obtaned for SFBC-SDBL-OTAS (8) scheme s shown n Table 4. 7 6 Bts per subcarrer 5 4 3 4 6 8 4 Subcarrer ndex Fg. 9. Bts per subcarrer vs. subcarrer ndex for SFBC-SDBL-OTAS (8) scheme: Case 3
An Improved Bt Loadng Technques for Enhanced Energy Effcency n.... 489.6.4 Energy per subcarrer...8.6.4. Group Index () 4 6 8 4 Subcarrer Index Fg.. Energy allocated per subcarrer vs. subcarrer ndex for SFBC-SDBL-OTAS (8) scheme: Case 3. Table 4. umber of bts assgned to the number of subcarrers n each group for SFBC-SDBL-OTAS (8) scheme: Case 3. Lower Index Upper Index B l b u b l u.77 4.43 39 4 5 7 8 4.43 8.87 88 5 6 4 8 3 3 8.87 6.5654 65 6 7 77 88 In Fg., average BER vs. SR (db) s compared between SISO-SDBL, SFBC-SDBL-AS, SFBC-SDBL-OTAS (4) and SFBC-SDBL- OTAS (8) systems for Case 3. SR s vared from to 4 db. The above mentoned systems reach a target BER of -4 at 3. db, 3.6 db,.4 db and 7.5 db respectvely. It can be observed that the proposed SFBC-SDBL-OTAS (8) scheme offers 4.7 db gan over the conventonal SISO-SDBL and 6. db gan over the SFBC- SDBL-AS systems.
49 K. V. Babu et al. - SISO-SDBL SFBC-SDBL-AS SFBC-SDBL-OTAS (4) SFBC-SDBL-OTAS (8) Average BER - -3-4 -5-6 5 5 5 3 35 4 SR (db) Fg.. Average BER vs. SR (db) comparson between the proposed schemes and the conventonal scheme: Case 3. In Fg., average spectral effcency (b/s/hz) vs. SR (db) s compared between SISO-SDBL, SFBC-SDBL-AS, SFBC-SDBL-OTAS (4) and SFBC- SDBL-OTAS (8) systems for Case 3. Snce the proposed algorthm s MA, the average spectral effcency ncreases up to certan SR and then t becomes constant. The proposed SFBC-SDBL-OTAS (8) scheme reaches a maxmum average spectral effcency of 5.778 b/s/hz approxmately at 6 db. SFBC- SDBL-OTAS (4), SFBC- SDBL-AS and SISO-SDBL schemes reach the maxmum value at db, db and 3 db respectvely. The proposed SFBC- SDBL-OTAS (8) scheme offers 4 db SR gan over the conventonal SISO- SDBL scheme. 6 Average spectral effcency (b/s/hz) 5.8 5.6 5.4 5. 5 4.8 SISO-SDBL SFBC-SDBL-AS SFBC-SDBL-OTAS (4) SFBC-SDBL-OTAS (8) 4.6 5 5 5 3 35 4 SR (db) Fg.. Average spectral effcency (b/s/hz) vs. SR (db) comparson between the proposed schemes and the conventonal scheme: Case 3.
An Improved Bt Loadng Technques for Enhanced Energy Effcency n.... 49 The smulatons are done wth the same parameters for Cases and. The average BER and average spectral effcency comparsons are only dsplayed for Cases and. The other results are not shown n ths paper. It s obvous that snce the proposed algorthm outperforms conventonal SDBL for Case 3, t wll naturally perform better for Case and Case. For Case, the target bts are fxed to. In Fg. 3, average BER vs. SR (db) s compared between SISO- SDBL, SFBC-SDBL-AS, SFBC-SDBL-OTAS (4) and SFBC-SDBL-OTAS (8) systems for Case. The above mentoned systems reach a target BER of -4 at db, 8 db, 6. db and. db respectvely. It can be observed that the proposed SFBC-SDBL-OTAS (8) scheme offers.9 db gan over the conventonal SISO- SDBL and 6.9 db gan over the SFBC-SDBL-AS systems. - SISO-SDBL SFBC-SDBL-AS SFBC-SDBL-OTAS (4) SFBC-SDBL-OTAS (8) Average BER - -3-4 -5-6 5 5 5 3 35 4 SR (db) Fg. 3. Average BER vs. SR (db) comparson between the proposed schemes and the conventonal scheme: Case. In Fg. 4, average spectral effcency (b/s/hz) vs. SR (db) s compared between SISO-SDBL, SFBC-SDBL-AS, SFBC-SDBL-OTAS (4) and SFBC- SDBL-OTAS (8) systems for Case. The proposed SFBC-SDBL-OTAS (8) scheme reaches a maxmum average spectral effcency of.389 b/s/hz approxmately at db. SFBC- SDBL-OTAS (4), SFBC-SDBL-AS and SISO- SDBL schemes reach the maxmum value at 5 db, 8 db and db respectvely. The proposed SFBC-SDBL-OTAS (8) scheme offers db SR gan over the conventonal SISO-SDBL scheme.
49 K. V. Babu et al..4 Average spectral effcency (b/s/hz).38.36.34.3.3.8.6 SISO-SDBL SFBC-SDBL-AS SFBC-SDBL-OTAS (4) SFBC-SDBL-OTAS (8).4 5 5 5 3 35 4 SR (db) Fg. 4. Average spectral effcency (b/s/hz) vs. SR (db) comparson between the proposed schemes and the conventonal scheme: Case. In Fg. 5, average BER vs. SR (db) s compared between SISO-SDBL, SFBC-SDBL-AS, SFBC-SDBL-OTAS (4) and SFBC-SDBL- OTAS (8) systems for Case. The above mentoned systems reach a target BER of -4 at.5 db, 5. db, 4 db and.8 db respectvely. It can be observed that the proposed SFBC-SDBL-OTAS (8) scheme offers.7 db gan over the conventonal SISO-SDBL and 4.3 db gan over the SFBC-SDBL-AS systems. - SISO-SDBL SFBC-SDBL-AS SFBC-SDBL-OTAS (4) SFBC-SDBL-OTAS (8) Average BER - -3-4 -5-6 5 5 5 3 35 4 SR (db) Fg. 5. Average BER vs. SR (db) comparson between the proposed schemes and the conventonal scheme: Case. In Fg. 6, average spectral effcency (b/s/hz) vs. SR (db) s compared between SISO-SDBL, SFBC-SDBL-AS, SFBC-SDBL-OTAS (4) and SFBC- SDBL-OTAS (8) systems for Case. The proposed SFBC-SDBL-OTAS (8) scheme reaches a maxmum average spectral effcency of 3.556 b/s/hz approxmately at 9 db. SFBC-SDBL-OTAS (4), SFBC-SDBL-AS and SISO- SDBL schemes reach the maxmum value at db, 4 db and db respectvely. The proposed SFBC-SDBL-OTAS (8) scheme offers db SR gan over the conventonal SISO-SDBL scheme.
An Improved Bt Loadng Technques for Enhanced Energy Effcency n.... 493 3.7 Average spectral effcency (b/s/hz) 3.6 3.5 3.4 3.3 3. 3. 3 SISO-SDBL SFBC-SDBL-AS SFBC-SDBL-OTAS (4) SFBC-SDBL-OTAS (8) 5 5 5 3 35 4 SR (db) Fg. 6. Average spectral effcency (b/s/hz) vs. SR (db) comparson between the proposed schemes and the conventonal scheme: Case. Average energy effcency of Case, and 3 for varous schemes s lsted n Table 5. It s very clear that the proposed SFBC-SDBL-OTAS (8) scheme offers better energy effcency for all the cases. Table 5. Average energy effcency (b/j) for the Cases, and 3 usng varous schemes. Schemes Case Case Case 3 SISO-SDBL.5 8.86 4.35 SFBC-SDBL-AS 68.6 8.87 7.97 SFBC-SDBL-OTAS (4) 3.3 37.49 9.78 SFBC-SDBL-OTAS (8) 33.5 75.75 65. 6. Concluson The MA-SFBC-SDBL-OTAS scheme s proposed n ths paper. The performance of the proposed algorthms s tested under varous channel condtons for all the three cases of target bts. For all channel condtons and for all three cases, the proposed scheme offer better spectral effcency, energy effcency and mproved BER performance than the conventonal SISO-SDBL scheme. The performance of the conventonal SDBL scheme s poor for Case 3. But after ncludng SFBC and antenna selecton schemes ts performance has been mproved. OTAS scheme allows the system to always choose best channels. The proposed scheme converts Case 3 problem to Case.
494 K. V. Babu et al. Ths algorthm needs just an teraton to converge whch reduces the complexty of system. Snce the algorthm offers better energy effcency, t s well suted for fxed rate wreless voce and vdeo applcatons. References. Das, S.S.; De Carvalho, E; and Prasad, R. (8). Performance analyss of OFDM systems wth adaptve sub carrer bandwdth. IEEE Transactons on Wreless Communcaton, 7(4), 7-.. Mao, G.; Hmayat,.; and L, G.Y. (). Effcent lnk adaptaton n frequency-selectve channels. IEEE Transactons on Communcatons, 58(), 545-554. 3. Feng, D.; Jang, C.; Lm, G.; Leonard, J.C.; Feng, G.; and L, G.Y. (3). A Survey of energy effcent wreless communcatons. IEEE Communcatons Surveys & Tutorals, 5(), 67-78. 4. Trved, Y..; and Chaturved, A.K. (). Performance analyss of multple nput sngle output systems usng transmt beamformng and antenna selecton wth delayed channel state nformaton at the transmtter. IET Communcatons, 5(6), 87 834. 5. Torab, M. (8). Adaptve modulaton for space-frequency block coded OFDM systems. Internatonal Journal of Electroncs and Communcatons, 6(4), 5-53. 6. Esfahan,.S.; and Afrasab, M. (7). Smple bt loadng algorthm for OFDM-based systems. IET Communcatons, (3), 3-36. 7. Rappaport, T.S. (). Wreless Communcatons, nd Edton, Prentce Hall PTR. 8. Hughes-Hartogs, D. (987,988,989). Ensemble modem structure for mperfect transmsson meda. U.S. Patents os. 4,679,7, 4,73,86, and 4833796. 9. Fscher, R. (996). A new loadng algorthm for dscrete multtone modulaton, IEEE Global Telecommuncatons Conference Globecom 96, London, UK, 74-78.. Chow, P.S.; Coff, J.M.; and Bngham, J.A.C. (9). A practcal dscrete mult tone transcever loadng algorthm for data transmsson over spectrally shaped channels. IEEE Transacton on Communcaton, 43(34), 773-775.. Babu, K.V.; Reddy, G.R; Bshakha, G.; and Sneha, D. (3). Improvement of qualty of servce (QoS) n MISO-OFDM systems usng superposton based adaptve modulaton (SPAM) and space frequency block codng (SFBC) technque. Internatonal Journal of Engneerng and Technology, 5(3), 8-5.. John, B.B.; Babu, K.V.; Reddy, G.R; and Balakrshna,. (4). A reduced MIMO (R-MIMO) system usng norm based approach (BA) for transmt antenna selecton (TAS). IEEE Internatonal Conference on Communcaton and Sgnal Processng ICCSP 4, Melmaruvathur, Inda, 47-43. 3. Babu, K.V.; Reddy, G.R; Balakrshna,.; John, B.B.; and Suraj, G. (4). Adaptve transmt dversty selecton (ATDS) based on STBC and SFBC for
An Improved Bt Loadng Technques for Enhanced Energy Effcency n.... 495 X MIMO OFDM systems. Internatonal Journal of Research n Engneerng and Technology, 3(7), 349-354. 4. Jung, J.; Kwon, B.; Park, H.; and Lm, J. (). Superposton based adaptve modulated space tme block codng for MIMO-OFDM systems. IEEE Communcatons Letters, 4(), 3-3. 5. Babu, K.V.; Reddy, G.R; and Utpal, S. (3). Spectral effcency mprovement n MIMO OFDM systems wth nterleave dvson multplexed (IDM) space tme block codng (STBC) scheme. Australan Journal of Basc and Appled Scences, 7(4), 77-83. 6. Gore, D.; and Paulraj, A. (). MIMO antenna subset selecton wth space tme codng. IEEE Transactons on Sgnal Processng, 5 (), 58-588. 7. Alamout, S.M. (998). A smple transmt dversty technque for wreless communcatons. IEEE Journal on Selected Areas n Communcaton, 6(8), 45-458. 8. Blum, R.S.; and Wnters, J.H. (). On optmum MIMO wth antenna selecton. IEEE Communcaton Letters, 6(8), 3-34