LLR Reliability Improvement for Multilayer Signals

Transcription

1 LLR Reliability mpovement fo Multilaye Signals C. Regueio, J. Baueco, J. Montalban, P. Angueia, J.L. Odiales, M. Velez Abstact The limitations and igidness of cuent use of spectum esouces have fosteed the design of new powe multiplexing techniques to impove fequency efficiency and flexibility. These techniques ae based on the simultaneous tansmission of diffeent sevices on the same channel with diffeent powe distibution. At the eceive side LLR pobability density functions have been so fa optimized fo single laye systems, so the eception quality of multilaye signals with the existing LLR algoithms is degaded. This pape poposes a new LLR pobability density function, including coections fo the multilaye influence, in ode to impove eception pefomance. Simulation esults ae also included to test the pefomance impovement in tems of the eceiving SNR theshold. ndex Tems LDM, Log-likelihood atio, LLR, multilaye, NOMA, PDF, simulations. NTRODUCTON he cuent adio spectum scacity has led to the Tdevelopment of new technologies that maximize the spectum efficiency. n this scenaio, the eutilization of the limited existing spectum to tansmit supeimposed sevices stacked in the same RF channel, though powe allocation multiplexing, is pesented as an inteesting solution. The Non- Othogonal Multiple Access (NOMA [] and the Layeed Division Multiplexing (LDM [], included in the new ATSC 3.0 candidate standad [3], ae new esouce shaing techniques based on the tansmission of two o moe diffeent independent signals supeimposed togethe at diffeent injection levels (L to fom a multilaye signal. n the eceive, consecutive signal cancellation pocesses must be applied to coectly decode the ovelaid signals [4]. Consideing, fo simplicity, a scenaio of two simultaneous signals, the eception of the most poweful signal, also called uppe laye (UL, is coupted by the supeimposed signal with lowe powe, also called lowe laye (LL. The eception of the LL signal, in contast, is not degaded as the UL cancellation is fist caied out and no othe signals ae pesent in the decoding pocess. These new scenaios ae possible due to the use of vey obust Low Density Paity Check Codes (LDPC [5][6] o tubocodes [7] in the UL. t is impotant to note that it is well known that soft decision decoding algoithms outpefom had decision decoding algoithms since they make use of eliability measues to gain knowledge of the tansmitted codewods [8]. Log-likelihood atio (LLR has been shown to The authos ae fom the Depatment of Communications Engineeing, Univesity of the Basque County (UPV/EHU, Alda. Uquijo s/n, Bilbao 4803, Spain. ( s: {cistina.egueio, jon.baueco, jon.montalban, pablo.angueia, juanluis.odiales, manuel.velez}@ehu.eus. be a vey efficient measue. n paticula, since LLR is the input to the soft input decode, knowledge of its pobability density function (PDF is equied. Gaussian modeling of the PDF has taditionally been used. t depends on the channel estimation, ρ, and the existing noise powe, N 0, as shown in ( [9], whee ( t,q t and (,Q epesent the tansmitted and eceived in-phase ( and quadatue (Q codewods, espectively. PDF t Q Qt (, /, N0 i Q t Qt e N0 n the absence of any othe noise souce, only Additive White Gaussian Noise (AWGN powe is consideed (N 0 =N AWGN. This equation fits well fo single laye systems but it does not take into account the LL influence on the UL decoding pocess in multilaye systems. n this pape a modification on the PDF of the LLR fo multilaye systems is poposed in Section. n addition, the theoetically defined fomula impact in the system pefomance is tested in Section. Finally, the main conclusions ae gatheed in Section V.. THEORETCAL STUDY The theoetical study pesented in this wok applies fo a two layes system; nevetheless it can be diectly extended to moe layes. The fist appoach pesented in the liteatue consides the LL as a white noise intefeence added to the extenal AWGN powe (N AWGN [0] []. Consideing that the multilaye (ML signal powe (P ML is the sum of the UL signal powe (P UL and LL signal powe (P LL, which is lowe in an njection level (L facto, the oveall noise powe to be consideed is defined in (: N ( 0 L /0 PML N AWGN PLL N AWGN ( / ( L Thus, the LLR PDF calculation in the decode was done based on ( taking N 0 as indicated in (. Howeve, the LL is an OFDM signal and its influence on the UL is diffeent fom that caused by the AWGN. Fo this eason, this coase appoximation of the LL signal as AWGN leads to degadation in the UL decoding pocess and, consequently, a new appoach should be consideed. The new poposal fo the LLR PDF in the UL decoding pocess is based on the study of PDF of the Q components of the eceived multilaye signal fo diffeent LL powes (depending on the specific L and diffeent AWGN powe

2 levels. By means of cuve-fitting, the PDF analytical solution in each case is calculated. This solution is based on the sum of as many Gaussian PDFs as LL constellation /Q samples ae. These Gaussian distibutions conside only the extenal noise powe level fo its vaiance (N 0 =N AWGN, and thei aveage values depend on the LL constellation and the L between the UL and LL of the multilaye signal. PDF fo the LLR in the decode based on the new appoach should be modified accoding to the following moe accuate fomula shown in (3. n this case, (,Q and ( ML,Q ML epesent the ML tansmitted and eceived Q codewods, espectively, as defined in (4, whee k spans each of the possible LL LL samples depending on the LL constellation. ML QML Q PDF NAWGN i ( / e (3 ML ML N ( ( ML ML ( ( UL UL 0 AWGN L/ 0 ( L / 0 0 k LL ( LL LL k LLk.(4 Fig. and Fig. show, in black dots, some example of the PDF of the in-phase component of the eceived multilaye signal fo a noise-fee (SNR = 30 db and a noisy (SNR = 0 db envionment, espectively. n blue, the esulting LLR PDF of a Gaussian distibution based in ( consideing that the LL is an additional Gaussian noise added to the existing AWGN as shown in (. n ed, the impoved LLR PDF of the eceived signal following (3 and (4 is shown. n these cases, the L between the UL and LL is 3 db and the LL uses a 6 QAM modulation (fou possible in-phase samples. As it can be seen, the new appoach (ed based in equation (3 and (4 is much moe accuate to the eceived signal PDF (black than the oiginal appoach (blue based in ( and (, especially fo high SNR. n low SNR envionments, the existing AWGN powe is highe than the LL powe and, consequently, the majo influence is due to the AWGN. n this situation, the poposed distibution fom equation (3 and (4 is simila to the Gaussian one, as it can be seen in Fig.. Fig. Example of the PDF of the eceived multilaye signal (black dots, the LLR PDF consideing the LL as AWGN (in blue and with the new appoach (in ed in a noise-fee envionment (SNR = 30 db. Fig. Example of the PDF of the eceived multilaye signal (black dots, the LLR PDF consideing the LL as AWGN (in blue and with the new appoach (in ed in a noisy envionment (SNR = 0 db.. SMULATON RESULTS n this section, seveal simulations have been caied out in ode to validate the new appoach fo the LLR PDF defined in (3 and (4, measuing the gain ove the use of ( and (. Fo this pupose, the value of the impovement gain, in tems of multilaye SNR thesholds fo coect UL decoding, is calculated with the two options. Fou diffeent studies have been caied out. Fist, the influence on the gain of the specific L between UL and LL is measued in Section.A. Next, the influence of the LL constellation is also studied in Section.B. Moeove, the specific UL constellation influence on the pefomance gain is also analyzed fo a two layes system in Section.C. Finally, the study is extended to a thee layes system in Section.D. A softwae implementation of an ATSC 3.0 multilaye system (LDM pofile [] has been used, supeimposing each laye with specific injection levels, afte being sepaately fomatted, encoded and modulated. Pefect time and fequency synchonizations ae assumed, and ideal channel estimation is consideed. n addition, the multilaye SNR is pefectly estimated as the aveage value pe FEC block. Thus, the minimum possible SNR thesholds ae defined as the SNR points satisfying the condition that the FEC Block Eo Rate (FBER at the output of the oute code is null. They ae numeically obtained by a bute-foce seach with a simulation step of 0.dB [3]. The evaluation has been caied out fo a Rayleigh channel model [4], which is widely used as efeence fo potable indoo eception in wieless communications, which is one of the potential taget of the UL sevice [5][6]. A. njection Level influence Table shows the powe distibution of the UL (P UL and LL (P LL signals in a nomalized multilaye signal fo diffeent L values. Fo testing puposes, five diffeent UL configuations have been consideed: QPSK modulation with 3/5, 4/5, 5/5, 6/5 and 7/5 code ate. The LL specific configuations ae detailed in each section. Table shows the theoetical SNR theshold fo a Rayleigh channel and the equivalent acceptable N AWGN (dbm of the nomalized UL configuations when no LL is consideed (SL pofile with P UL = 0 dbm.

3 3 TABLE POWER DSTRBUTON ON A TWO LAYERS SGNAL DEPENDNG ON THE L L (db P UL (dbm P LL (dbm TABLE UL CONFGURATON SNR THRESHOLD (DB N SNGLE LAYER PROFLE UL P UL (dbm SNR (db N AWGN (dbm QPSK 3/ QPSK 4/ (Nomalized QPSK 5/ SL pofile QPSK 6/ QPSK 7/ n a multilaye system, the UL suffes fom two diffeent effects. On the one hand, the P UL deceases with lowe values of L between UL and LL signals. Consequently, the acceptable N AWGN fo UL coect decoding at the theshold situation is also deceased. On the othe hand, as it has been stated befoe, the LL acts as an additional intefeence powe that cannot be eliminated. Consideing in a fist appoach the LL as white noise, it also educes the acceptable N AWGN at the theshold situation in ode to maintain the same SNR theshold fo UL. n consequence the theoetical acceptable N AWGN at the theshold situation fo UL coect decoding depends on the UL theshold SNR in SL pofile and on the L between both layes. Fig. 3 shows the theoetical N AWGN depending on the L between UL and LL fo the five consideed UL configuations when the LL is consideed as an additional white noise. The P LL fo its L value is also shown. Fig.3 shows that the N AWGN needed fo eaching the coect eception theshold situation is almost constant afte a cetain value of L. Fom this L value, the LL powe influence on the SNR theshold is almost negligible and, consequently, the new poposal based on using (3 and (4 does not povide any gain. This limit L value depends on the specific obustness of the UL configuation, as the less obust the configuation is, the influence of the LL on the UL becomes negligible fo highe L values. Fom Fig. 3, it can be stated that the majo LL influence fo the consideed UL configuations is with L of up to 5 db. Fom this eason, in ode to establish the gain of using the new appoach based on (3 and (4 depending on the L, some compute simulations have been caied out fo L anging fom db to 5 db. Fig. 3 Acceptable N AWGN (db fo UL coect decoding fo diffeent Ls. Table shows the SNR theshold fo coect UL decoding when using the oiginal appoach (equations ( and ( and the new appoach (equations (3 and (4 fo the consideed UL configuations. By this way, the eal impovement gain of using the new suggested appoach can be measued. Besides, the theoetical SNR theshold consideing the acceptable N AWGN fom Fig. 3 is also shown in ode to make easie the undestanding of the gain value. n this case, the LL signal is a 6 NU-QAM /5. The (* means that the decoding pocess is incoect fo a multilaye SNR of up to 0 db, so no impovement gain can be assessed, showing an Undefined value. TABLE SNR THRESHOLD AND MPROVEMENT GAN (DB FOR DFFERENT L UL L Theoy (+( (3+(4 Gain (db QPSK 3/ QPSK 4/ QPSK 5/ (*.6 Undefined QPSK 6/ (* (* 7.8 Undefined (* (* 0.8 Undefined QPSK 7/ As it can be stated fom Table, the UL decoding using the new appoach in (3 and (4 does not mean degadation in any of the analyzed cases. n fact, its use almost always means gain fo the consideed L values. The gain value inceases with lowe L values because, as expected, the effect of the LL

4 4 on the UL is highe fo low L (db and the influence of the new LLR PDF fomula is moe significant. Besides, the gain impovement fo the same L is highe fo less obust configuations, as the acceptable N AWGN is lowe, (see Fig. 3 and consequently, the existing P LL is compaatively highe than the N AWGN. Unde that cicumstance, the use of the new appoach based on (3 and (4 is moe noticeable with less obust configuations and the gain value inceases significantly. Moeove, if the new expession to calculate LLRs PDF (equations (3 and (4 is consideed, the SNR theshold fo coect UL detection is close to the theoetical one. n fact, the pefomance of the new appoach sometimes ovetakes the theoetical pefomance based on consideing the LL as AWGN. This fact happens when the P LL is highe than the theoetical acceptable N AWGN, as it can be seen in Fig. 3, so the new LLR PDF fomula optimizes its pefomance. Fig. 4 shows the theoetical acceptable N AWGN, the value obtained with ( and ( and the one obtained using (3 and (4, fo all the UL tested configuations and L values. As it can be seen in Fig. 4, the acceptable N AWGN obtained with the oiginal LLR PDF fomula based on ( and ( diffes fom the theoetical value. The diffeence inceases fo low L and less obust configuations, when the LL has a majo impact. n this situation, the new LLR PDF fomula based on (3 and (4 especially inceases its pefomance, getting moe accuate N AWGN values. Fig. 4 SNR theshold in theoy, with ( + ( and with (3 + (4 B. LL Constellation influence As the new poposal based on (3 and (4 depends on the numbe of /Q samples of the LL constellation, the impovement gain ove using the oiginal appoach based on ( and ( depending on the specific LL constellation size has also been tested. This analysis has been caied out fo 6 NU- QAM /5 (4 bits, 64 NU-QAM /5 (6 bits, 56 NU- QAM /5 (8 bits and 04 NU-QAM /5 (0 bits [7]. The consideed L value anges fom to 5 db. Table V shows the impovement gain of using the new LLR PDF fomula (equations (3 and (4 fo QPSK 3/5 UL configuation. Simila esults have been also obtained fo the est consideed UL configuations, so they ae not included in Table V. TABLE V SNR THRESHOLD AND MPROVEMENT GAN (DB FOR DFFERENT L AND LL CONSTELLATONS (UL: QPSK 3/5 LL L (+( (3+(4 Gain (db NU-QAM NU-QAM NU-QAM NU-QAM As it can be stated fom Table V, although the definition of the new appoach in (3 and (4 depends on the specific LL constellation, thee ae no high diffeences (always lowe than 0. db on the gain values fo diffeent LL constellations. This is because the specific LL constellation is consideed in the shape of the LLR PDF fomula in each case. C. UL constellation influence The UL constellation influence on the impovement gain of using (3 and (4 instead of ( and ( has also been tested. Diffeent UL constellations sizes have been consideed: QPSK ( bits, 6 NU-QAM (4 bits and 64 NU-QAM (6 bits. The specific code-ate in each case is the one that means a simila SNR theshold fo coect decoding in a Rayleigh channel model when no LL is pesent (SL pofile. By this way, the influence of the UL specific constellation is studied with no influence of the diffeent obustness level. Moeove, as it has been demonstated in subsection A, the specific LL constellation has vey low influence on the impovement gain value, so the last LL constellation fom the pevious study (04 NU-QAM has been consideed fo the simulations. n this case, the L also anges fom to 5 db. Table V includes the SNR theshold when using the oiginal appoach (( and ( and the new LLR PDF ((3 and (4. By this way, the gain impovement is assessed fo diffeent UL constellations. Besides, the theoetical SNR theshold has also been included. The (* means that the decoding pocess is incoect fo a multilaye SNR of up to 0 db, so no gain can be assessed, showing an Undefined value. As it can be seen, fo simila obust UL configuations the gain is highe fo highe ode modulations. This is because the distance between consecutive Q points is smalle in highe constellations. Fo this eason, the LL inceases its influence on the UL fo the same L. Consequently, the

5 5 impovement because of the use of the new LLR PDF appoach ((3 and (4 instead of the oiginal one (( and ( inceases with the UL constellation ode. TABLE V SNR THRESHOLD AND MPROVEMENT GAN (DB FOR DFFERENT L AND UL CONSTELLATONS UL L Theoy (+( (3+(4 Gain (db 6.8 (*.6 Undefined QPSK 6/ (* 4.8 Undefined 8.7 (* 8.4 Undefined 6 NU-QAM 3/ (* 3.0 Undefined 9. (* 8.5 Undefined 64 NU-QAM / Howeve, the SNR theshold using (3 and (4 is simila fo the thee UL constellations odes, as the tested configuations have also simila pefomance when no LL is pesent. The diffeences in the gain value ae due to the diffeent SNR theshold when ( and ( ae consideed. This value inceases with the constellation ode esulting in highe degadation fo UL highe constellations. Theefoe, the UL highe ode constellations suffe moe degadation because of the pesence of a LL when ( and ( ae consideed. Howeve, the impovement of using (3 and (4 is also highe fo high ode UL constellations in ode to coect the specific degadation in each case. D. Thee Layes pefomance f moe than two layes ae tansmitted in the same adiofequency channel at once, the influence of all of them should be taken into account [8]. f thee layes ae consideed, the UL is influenced by LL with an L and LL constellation; and LL with an L and LL constellation. n this case, the LLR PDF fomula defined in (3 is completely valid but with the new definition in (5 of tansmitted and eceived Q codewods of the multilaye signal, (,Q and ( ML,Q ML, espectively. ( ( ML ML ( ( UL UL 0 k L /0 0 ( L /0 ( k 0 k L /0 k 0 ( LL L /0 ( LL LL k LL k,(5 whee k spans each of the possible samples depending on the LL constellation and k spans each of the possible LL LL samples depending on the LL constellation n ode to measue the impovement gain of using the new LLR PDF appoach ((3 and (5 instead of the oiginal one (( and (, the used ATSC 3.0 multilaye system has been adapted to the simultaneous tansmission of thee layes instead of only two as defined in the standad. Table V shows the powe distibution of the UL (P UL and LLs (P and P LL signals in a nomalized multilaye signal fo diffeent L and L values anging between and 5 db. TABLE V POWER DSTRBUTON ON A THREE LAYERS SGNAL DEPENDNG ON THE L L (db L (db P UL (dbm P (dbm P LL (dbm Equally to the two layes system, the acceptable N AWGN at the theshold situation depends on the L and L and the UL configuation obustness as it can be seen in Fig. 5. Consequently, the theoetical SNR also depends on L and L. Fig. 5 Acceptable N AWGN (db fo UL coect decoding fo diffeent Ls in a thee layes system As it can be seen in Fig. 5, the influence of the L and L on the most obust configuations is low, especially fo high L values. Howeve, less obust UL configuations (QPSK 6/5 o 7/5 ae moe influenced by the specific L and, in geneal, cannot be coectly decoded with two layes, especially if both LL and LL ae vey low. Taking eveything into consideation, in this case, a QPSK 4/5 UL has been analyzed as it is obust enough to be coectly decoded with LLs inseted with low L and it is no so obust to be able to detect the influence of the LLs ove the existing AWGN in the theshold situation. As it has been demonstated in section -A, the specific LL constellation has vey low influence on the impovement gain, so only two LL constellations have been consideed (6 NU-QAM fo LL and 64 NU-QAM fo LL.

6 6 Table V shows the impovement gain of using (3 and (5 instead of ( and ( with diffeent L and L values. The (* means that the decoding pocess is incoect fo SNR lowe than 0 db, and consequently the gain stays Undefined. TABLE V SNR THRESHOLD AND MPROVEMENT GAN (DB FOR DFFERENT L AND L N A THREE LAYERS SYSTEM (UL: QPSK 4/5 L (db L (db Theoy (+( (3+(5 Gain (db (* (* (* Undefined.8 (* 6.3 Undefined (* 0.3 Undefined (* 8.6 Undefined (* 7.6 Undefined 8.4 (* 9.7 Undefined (* 7.6 Undefined The esults follow the same tendency than with only one LL. On the one hand, the highe the L and L ae, the lowe the impovement is as the LLs have lowe influence on the UL. n fact, when L is vey low, the use of (3 and (5 enables an UL coect decoding at the eceive wheeas it is not possible with the oiginal appoach based on ( and (, obtaining a vey high impovement gain. Moeove, although the gain cannot be exactly estimated fo vey low L (such as o db, a minimum gain can be deduced with value fom 3.7 db up to.4 db depending on the specific L value. The LL decoding impovement is not analyzed as once the UL has been cancelled, a two layes system is pesent and the esults will follow the same tendency than in the pevious subsections. V. CONCLUSONS This pape has poposed a new LLR PDF to be used in the eception of signals using multilaye multiplexing techniques. This new fomula takes into account the degadation of a supeimposed laye on the desied signal though a change in the expession of the LLR PDF fomula. The new expession, defined in (3 and (4, depends on the injection level between layes and the specific lowe laye constellation. Seveal compute simulations have been caied out with an ATSC 3.0 tansmission/eception platfom, using the Laye Division Multiplexing (LDM pofile. As a esult, the multilaye SNR theshold fo the coect uppe laye decoding pefomance is neve degaded and can be impoved in up to 4.5 db, depending mainly on the signal obustness, the specific signal constellation and the injection level. Besides, it has also been confimed that the highe SNR equiement the signal has, the highe the new poposal gain, as the AWGN pesent in the theshold situation is lowe and the lowe laye has moe impact on the pefomance. Fo this eason, these esults can be extended fo moe challenging scenaios esulting in highe gain values. Moeove, it has been also poved that the new LLR PDF expession can be extended to moe than two layes with even highe pefomance gain. Thus, the implementation of this new algoithm means that the eceives could coectly decode the uppe laye with lowe signal levels, inceasing consequently the coveage aea. ACKNOWLEDGMENT This wok has been financially suppoted in pat by the Univesity of the Basque County (UF /30, in pat by the Basque Govenment (T and PREDOC pogam and in pat by the Spanish Ministy of Economy and Competitiveness (unde Pojects HEDYT-GBB, TEC0-3330, and 5G-NewBROs, TEC P MNECO/FEDER and the Euopean Regional Development Fund, ERDF REFERENCES [] Y. Saito, et al., Non-Othogonal Multiple Access (NOMA fo Cellula Futue Radio Access, EEE Vehicula Technology Confeence (VTC, pp. -5, June 03. [] L. Zhang, et al., Channel capacity distibution of Laye-Division- Multiplexing system fo next geneation digital boadcasting tansmission. EEE ntenational Symposium on Boadband Multimedia Systems and Boadcasting (BMSB 04, pp. -6, June 04. [3] ATSC (Advanced Television System Committee, A/3: ATSC Poposed Standad Physical Laye Potocol, May 06. [4] J. Montalban, et al, Cloud Tansmission fequency domain cancellation, EEE ntenational Symposium on Boadband Multimedia Systems and Boadcasting (BMSB 03, pp. -4, June 03. [5] S-. Pak; H-M. Kim; Y. Wu; Jeongchang Kim, "A Newly Designed Quate-Rate QC-LDPC Code fo the Cloud Tansmission System," EEE Tans. on Boadcasting, vol.59, no., pp.55-59, Mach 03. [6] Y Li, et al., Rate-Compatible LDPC-R-S Poduct Codes Based on Rapto-like LDPC Codes, EEE ntenational Symposium on Boadband Multimedia Systems and Boadcasting 03 (BMSB03, pp. -6, June 03. [7] N. Souto, J.C. Silva, F. Cecas, Low Rate Tubo Codes based on Nonlinea Cyclic Codes, EEE ntenational Confeence on Communications, vol., pp , June 004. [8] R. Jose and A. Pe, Analysis of Had Decision and Soft Decision Decoding Algoithms of LDPC Codes in AWGN, EEE ntenational Advance Computing Confeence (ACC 05, pp , June 05. [9] V. Guuswami, teative Decoding of Low-Density Paity Check Codes (An ntoductoy Suvey, Euopean Association fo Theoetical Compute Science (EATCS, ssue: 90, Octobe 006 [0] J. Montalbán, et al., Cloud tansmission: System pefomance and application scenaios, EEE Tansactions on Boadcasting, vol.60, no., pp.70-84, June 04. [] J. Montalban, et al., Pefomance Study of Layeed Division Multiplexing Based on SDR Platfom, EEE Tansactions on Boadcasting, vol.6, no.3, pp , June 05. [] C. Regueio, et al., ATSC 3.0 nteleaves nfluence in Reception Pefomance, EEE ntenational Symposium on Boadband Multimedia Systems and Boadcasting 06 (BMSB06, pp. -4, June 06. [3] C. Regueio, et al., Field Tials-Based Planning Paametes fo DVB- T ndoo Reception, EEE Tansactions on Boadcasting, vol.6, no., pp.5-6, Febuay 05.

7 [4] Recommendation TU-R; Planning citeia, including potection atios, fo digital teestial television sevices in the VHF/UHF bands," BT.368-; Febuay 05. [5] C. Regueio, et al, LDM Coe Sevices Pefomance in ATSC 3.0, EEE Tansactions on Boadcasting, vol.6, no., pp.44-5, Januay 06. [6] C. Regueio, et al. SHVC and LDM techniques fo HD/UHD TV indoo eception, EEE ntenational Symposium on Boadband Multimedia Systems and Boadcasting 05 (BMSB05, pp. -6, June 05. [7] L. Michael and D. Gómez-Baqueo, Bit-nteleaved Coded Modulation (BCM fo ATSC 3.0, EEE Tansactions on Boadcasting, vol.6, no., pp.8-88, Januay 06. [8] J. Montalban, Asynchonous N-Layeed Division Multiplexing (N- LDM, EEE ntenational Symposium on Boadband Multimedia Systems and Boadcasting 06 (BMSB06, pp- -8, June 06. 7