Forward Link Interference Mitigation in Mobile Interactive Satellite Systems

F orw ard Link In terference Mitigatio n in Mobile In teractiv e S atellite Systems P . Henarejos ∗ , M. A. Vázquez † and G. Co cco ‡ Centr e T e cnolò gic de T ele c omunic acions de Catalunya, 08860 Castel ldefels, Bar c elona, Sp ain A. P érez-Neira § Universitat Politè cnic a de Catalunya, 08034 Bar c elona, Sp ain W e presen t the results of the p erformance ev aluation of p olarization-time co ding and soft in terference cancellation in m ulti-b eam satellite systems aected b y co-c hannel in terference in r ealistic setups. The standar d of B roadband Global Area Net w ork service (BGAN) has b een considered as reference for the ph ysical la y er and realistic in terference and c hannel mo dels ha v e b een adopted. T he w ork has b een carried out in the framew ork of t he Next Generation W a v eform for Increased Sp ectral Eciency (NGWISE) pro ject founded b y the Europ ean Space Agency (ESA). I. In t ro duction In teractiv e satellite systems are a k ey comm un i cation solution with a h u g e p oten tial mark et. P ossible applications range from the pr o vision of data connectivit y to areas where cellular connection is n o t protable (e.g., rural) or infeasible (e.g., maritime and aircraft sc enarios) to bac king up in emergency situations, esp eciall y when in v olving large geographical areas. F ar from b eing limited to broadcast t ransmi s sions, the ab o v e men tioned applications rely on m ultic ast and m ultiple unicast c onnections. In this framew ork m ulti-b eam satelli t e s can pro v i d e a man y-folds increase in sp ectral e ciency with resp ect to global b eam satellites, esp ecially in case of arc hitectures with a lo w frequency reuse factor. An ev en higher throughput can b e pro vided, in principle , b y lev erage on p olarization reusing on adjacen t b eams and transmission on b oth p olariza t i ons within a single b eam. Ho w ev er, the adoption o f aggressiv e fre- quency/p olarization reuse sc hemes implies an increase of in tra-system in terference due to satellite an tenna side-lob es, lo w directivit y o f user terminal (UT) an tennas and p olarization mi smatc h due to an tennas imp er- fections and atmospheric propagation. 1 Moreo v er, mobile terminals are also k e en to propagation impairmen t suc h as shado wing and fading, while the large propagation dela y t y pic al of satellite sy stems (esp eciall y GEO) prev en ts the a v ailabili t y of c hannel state information (CSIT). Time div ersit y is largely exploited in to da y's in teractiv e m obile systems st andards to o v ercome c hannel impairmen ts in mobile broadcast systems. Ho w e v er, new div ersit y tec hniques ha v e recen tly gained i n terest. Suc h tec hniques are based on p olarization and spatial div ersit y t hat al lo w to apply m ultiple input-m ultiple output (MIMO) tec hniques suc h as preco ding and p olarizati on-time co des. In terference cance llation tec hniques 2 are also a p oten tial solution that i s curren tly b eing lo ok ed at in b oth the forw ard and the rev erse link. Dual p olarization transmission has b een ev aluated for the mobil e broadcast scenario in 3 with promising results. The join t eect of outdated CSIT and the time v ariabilit y of c hannel mak es v ery dicult the use of linear preco ding as it w as applied in previous w orks. 4 , 5 Apart from an increase in sy stem div ersit y , the dual p olarization transmission can pro vide an increase in sp ectral eciency esp ecially in case of lo w cross-p olar in terference. ∗ Researc h Engineer, Comm un i ca ti o n Systems, p ol.henarejos@cttc.es † Researc h Engineer, Comm un i ca ti o n Systems, ma v azquez@cttc.es ‡ P ost-do ctorate Researc her, Comm unication Systems, gco cco@cttc.es § Professor at Signal Theory and Comm unications Departmen t at UPC and Senior Researc her at CTTC, ana.isab el.p erez@up c.edu 1 of 11 American Institute of A e ronautics and Astronautics In t he presen t pap er presen t part of the r e sults w e obtained within the Next Generation W a v eform for Increased Sp ectral Eci ency (NGWISE) pro ject founded b y the Europ ean Space Agency (ESA). 6 More sp ecicall y , w e ev aluate t he impact of dual p olarization transmission in terms of throughput in presence of co- c hannel in terference i n a m ulti-b eam satellite system and ev aluate the p ossibilit y of applying soft in terference cancellation (SIC) in in terference-lim ited setups. A realistic c hannel mo del is adopted and dieren t scenarios are considered, namely m ar i time and terrestrial. The standard adopted in Broadband Glo b a l Area Net w ork service (BGAN) 7 is used as reference standard for the ph ysical la y er. In the rest of the pap er w e will r e fer to 7 as BGAN st a n da r d for simplicit y . Our results sho w that that a higher sp ectral e ciency can b e ac hiev ed through the considered tec hniques, whic h ma y lead to an increased o v erall sy stem throughput. W e also sho w that for a four-color frequency reuse sc heme soft in terference cancellation pro vides limited gain in terms of blo c k e r ror rate due to the lo w relativ e p o w er and high n um b e r of in terferers t hat can b e assimilated to Gaussian noise. I I. System Mo del Let us consider the forw ard link of a m ulti-b eam geostationary satellite comm unication system. Co- c hannel in terference among adjacen t b eams is mitigated through f re q uenc y reusing. A four colors frequency reuse sc heme is considered in the follo w i ng. A dual p olarization transmission is assumed, i.e., the satellite and the user termi n a l an tennas transmits and receiv e o v er t w o (almost-)orthogonal p olarizatio n s, resp ectiv ely . The receiv ed sig n a l at the user terminal can b e expressed as: Y = √ P HBC ( s ) + HBJ + N (1) where Y ∈ C 2 × 2 represen ts the recei v ed signal in t w o time i n stan ts from the t w o p olarizations, P is the transmitted p o w er, H ∈ C 2 × 2 is the c hannel matrix, the distribution of whic h dep e nds on the scenario, B ∈ R 2 × 2 is a matrix that accoun ts for the an tennas c haracteristics in term s of co-p olar, cross-p olar gains and co-c hannel in t e rf e rence rejection while C ( s ) is the p olarization-time co de, whic h dep ends on the sym b ol v ector s ∈ C M × 1 , ha ving bl o c k length M . W e assume a general complex sym b ol mapping. The in ter-b eam in terference i s mo delle d through the matrix J ∈ C 2 × 2 . Thermal noise is tak en in to accoun t through the term N ∈ C 2 × 2 whose en tr i es are zero mean Gaussian r andom v ariables with v ariance N . As is often the case in terrestrial comm unications, also in the satell ite con te x t the use of M IMO tec h ni q ue s is com bined with c hannel co ding leading to what is called a M IMO-BICM mo dulation. Figure 1 depic ts the general b l o c k diagram of this sc heme. The b l o c k Π and Π − 1 in the picture represen t the co ded bit in terle a v er and dein terlea v er, resp ectiv ely . Figure 1. Equiv alen t MIMO-BICM sc heme of the consider ed dual p olariz a tion s e t u p. F or this scenario, the main c hallenges to b e fac ed are t w o. First, the MIMO transmission sc heme (i .e. ho w sym b ols are transmitted b y the t w o p o larizations) m u st b e decided. Secondly , t he MIMO demo dulator (i.e. ho w sym b ols from the t w o p olarizations are detected) needs to b e also obtained. F or b oth the reference signal (i.e., the signal that is to b e receiv ed b y the user t e rminal) and the in terfering signals the B GAN standard 7 is adopted. The BGAN standard, curren tly under denition, is designe d to supp ort b oth v oice and broadband data services in a wide range of scenario suc h as maritime and land mobile. The c hannel co de adopted in the BGAN standard is a turb o co de with sev eral p os sibl e congurations in terms of co de rate and co d e w ord length. The p ossible com binations of c hannel co de parame t e rs , mo dulation (QPSK, 16QAM), sym b ol rate and other ph ysical la y er c haracteristics are dened b y the b earers. I I I. P olarization-Time T ransmission and Reception S c hemes In the presen t section w e consider sev eral p ossible MI MO mo dulation and dem o dulation solutions pre- viously prop osed in literature that are suited for application in the considered setup. Out of t he m, one mo dulation and one demo dulation sc heme are sele cted based on practi cal as w ell as theoretical considera- tions. 2 of 11 American Institute of A e ronautics and Astronautics I I I.A. MIMO T ransmission Sc heme • Alamouti : The ob jectiv e of this p olarization-ti me co de is to obtain the m ax i m um div ersit y gain b y constructiv ely adding up the c hannel gains from the t w o p olarizations whil e rejecting the in ter-sym b ol in terference (ISI). 8 In this c ase M = 2 and the co ding matrix is C ala ( s ) = s 1 − s ∗ 1 s 2 s ∗ 2 ! . (2) The Maxim um-Lik eliho o d (ML) detection for this sc heme is kno wn to ha v e lo w complexit y , as it reduces to to a matrix m ultiplication and a set of com par i sons . After optimally deco ding this p olarization-time co de, the sym b ol obtains a div ersit y gai n of G d = k HB k 2 = | h 11 | 2 b 2 11 + | h 22 | 2 b 2 22 , (3) where b 11 and b 22 are the diagonal elemen ts of B . Note that this equiv alen t SISO gain is obtained when the ML detecti on is used. • P olarization Multiple xing : P olarization m ultiplexing obtains the full m ultiplexing gain (i.e. in t w o time instan ts 4 sym b ols are transmitted, 1 for eac h c hannel use of eac h an tenna). 9 F or this case M = 4 and the co ding matrix is C m ul ( s ) = s 1 s 3 s 2 s 4 ! . (4) • Golden Co de : t hi s is a full div ersit y tec hnique that still pro vides som e co ding gain. 10 The co ding matrix is constr uc ted as follo ws C gol ( s ) = s 1 + αs 2 s 3 + αs 4 i ( s 3 + β s 4 ) s 1 + β s 2 ! , (5) where α = 1+ √ 5 2 and β = 1 − √ 5 2 . I I I.B. MIMO-BICM Demo dulators A lo w complexit y detector for the MIMO sc hemes presen ted ab o v e is the hard decision ML detector. The ML decision rule consists in solving the follo wi n g optimization problem: arg min ˆ s k y − HC ( ˆ s ) k 2 2 . (6) Note that the detected sym b ol ˆ s is obtained via a hard decision. Ho w ev er, a c hannel co de is u sual ly included in all MIMO sc hemes, and th us the MIMO de mo dulator should output the log-lik eliho o d ratios (LLR) for the co ded sym b ols that are to b e passed t o the deco der 11 , 12 . The LLR for the l -th co ded sym b ol is Λ l = log p ( c l = 1 | y , H ) p ( c l = 0 | y , H ) , (7) where p ( c l | y , H ) is the probabilit y mass function of the c o ded bits conditi oned on the c hannel output y and the c hannel matrix H . In order to reduce the complexit y the log-sum appro ximation can b e applied with little loss in terms of p erformance. 11 In the follo wing w e consider dieren t sc hemes that aim at decre as i ng the demo dulator complexit y . In order to reduce the n um b er of demo dulators to test, w e refer to the recommendations in 13 where sev eral demo dul ator s w ere studied considering the m utual information as a me asu re men t. In 13 it w as sho wn that for lo w data rates the soft-minim um mean square error (MMSE) 14 , 15 demo dulator outp erforms the other designs. As an extension of this tec hnique w e prop ose the soft v ersion of the M MSE-SIC receiv er. In the follo wing brie  y describ e suc h sc hemes as w ell as th e optimal solution for the unco ded c ase. • Soft-MMSE : this is a linea r equalize r obtained through the minimization of the mean sq uare error (MSE). Its expr e ss i on is G MMSE =  H H H + σ 2 I  − 1 H H . (8) 3 of 11 American Institute of A e ronautics and Astronautics • Soft-MMSE-SIC : this detector is the soft v ersion of the V-BLAST demo dulator. This tec hnique iterativ ely deco des and subtracts the ISI. In order to adapt this to our setup w e substitute the hard decision of the i n terference with a soft one. The resulting algorithm is the follo wing: 1. Find the least faded p olarizati on: k i = min k g k (9) where g k , k = 1 , 2 are the comp onen ts of v ector g = diag   HH H  − 1  . 2. Obtain the MMSE estimate of signal s k i transmitted in the p olarization k i : ˆ s k i = [ G MMSE y n ] k i . (10) 3. Subtract in terference due to this sym b ol from y y n = y n − ˆ s k i [ H ] k i , (11) where y n is the receiv e d signal at time instan t n . 4. Remo v e k i th comp onen t from y n and the k i th column of H . 5. Rep eat 1-4 for the other p olarizati on. I I I.C. Syst em Design W e c hose one enco ding/deco ding sc heme to b e applied in our scenario among those presen ted so far. The c hoice has b een a trade o b et w een the or e tical considerations and practical constrain ts either dictated b y the BGAN standard or b y other practical considerations suc h as complexit y at the user terminal. • T ransmission sc heme: p olariz ation m ultiplexing. It w as sho wn in previous ESA pro jects that this metho d outp erforms the A lamouti one when considering a giv en transmit p o w er. Another p o s sibl e c hoice w oul d b e to use Gol den co des that, according to some preliminary results w e obtained, sho ws an impro v e of around 1dB in FER with resp ect to Alamouti. Ho w ev er the Golden co des imply an increase in computational complexit y as 4 sym b ols are mixed together in t w o time instan ts, whic h mak es the detection more computationally dem anding. Keeping in mind that in our study case lo w complexit y at the receiv er is an asset, w e considered that t he enhancemen t in terms of FER do es not justify the incre ase in complexit y . • Detection sc h e me: Soft-MMSE-SIC. As describ ed in 13 this sc heme outp erforms the other demo dulators and presen ts a go o d p erformance in the higher sp e ctral eciency region. As a matter of facts it can b e observ ed in gures 2 and 3 that the soft-MMSE dem o dulator p erforms slig h tly b etter in lo w data rates while at higher data rates it c an b e observ ed that the Soft-MMSE-SIC p erforms b etter. IV. Soft In t erference Can ce l ation The div ersit y/m ultiplexing gains p oten tially deliv ered b y p olarization-time sc hemes ma y s ue r from the co-c hannel i n terference coming fr o m other b eams dep ending on th e in terference strength. In rst appro xima- tion (more accurate if the n um b er of in terferers is large) suc h in terference can b e assimilated t o a bac kground noise whic h can not b e dealt with using the MI MO tec hniques pr e sen ted in S e ction I I I . In terference can- celation ma y he lp in suc h case. A rst classication of in terfere n c e cance lation metho ds can b e done b y distinguishing hard (HIC) and soft in terference cancelation (SIC). In HIC one of the signals (usually the strongest one) is deco ded treating the others as noise and than subtracted from the r e ceiv ed w a v eform. Suc h sc heme is relativ ely simple but has the dra wb a c k that, if the signal to b e subtracted is not dec o ded correctly , error propagation can sev erely limit the p erfor m ance of the system. SIC metho ds consist of a soft estimation of eac h of t he transmitted signals follo w ed b y a deco ding phase in whic h suc h estimation is tak en in to accoun t b y the deco der. Examples can b e found in, 16 17 and. 18 In the follo wing w e consider the iterati v e SIC sc heme depicted in Fig. 4 for the case of t w o receiv e d signals (one reference signal and one in terferer) in a SISO c hannel. In suc h sc heme the receiv ed w a v eform is fed to the soft e s ti mator, whic h p erforms detection and e s tim ates the transmitted c hannel sym b ols for b oth signals. The estimation is p erformed using a tur b o deco der that 4 of 11 American Institute of A e ronautics and Astronautics Figure 2. BLER for Soft-MMSE receiv er at cen ter of co v erage for maritime scenario. Figure 3. BLER for Soft-MMSE-SIC receiv er at cen ter of co v erage for maritime scenario. Figure 4. Iterativ e SIC sc heme. 5 of 11 American Institute of A e ronautics and Astronautics b een mo died to output soft estim ates of de considered signal. The detecti on/estimation is ite r a t e d a n um b er N iter of times, after whic h a decision is tak en on the sym b ols of b oth signals. The same metho d can easily b e extended to the dual p olarizati on case esp eci ally in case of high c ros s p olari zation rejecti on. V. Numerical Results In this section w e prese n t the p erformance ev aluation results for t he sele cted M IMO sc heme in t w o scenarios, namely maritime and land mobil e satellite with in termediate tree shado wi n g (LM S-ITS). The BGAN standard, op erating in L-band has b een adopted a refere n c e for the ph ys i cal la y er. W e consider the maritime scenario rst. The c hannel mo del is describ ed i n the follo wing table. F ast fading Rician K Rician factor 10 dB Doppler shift 2 Hz # taps 1 T a b le 1. Channel parameters W e fo cus on b earers t yp es with sym b ol rate 33 . 6 Ksym b ols/s. M or e sp ecically , w e fo cus o n the BGAN F80T1Q4B b earer t yp es c haracterized b y QPSK mo dulation. The co de rate for the dieren t F80T1Q4B sub-b earers is describ ed in the follo wing table. T abl e 2. Co de Rate of F80T1Q4B b earer t yp es Bearer Name Co d i n g Rate Data Rate (Kbps) F80T1Q4B-L8 0.34 21.6 F80T1Q4B-L7 0.40 25.6 F80T1Q4B-L6 0.48 30.4 F80T1Q4B-L5 0.55 35.2 F80T1Q4B-L4 0.63 40.0 F80T1Q4B-L3 0.70 44.8 F80T1Q4B-L2 0.77 49.2 F80T1Q4B-L1 0.83 52.8 F80T1Q4B-R 0.87 55.6 Dep ending on the geographical lo cation of the u se r terminal (cen ter/edge of co v erage, cen ter/edge of b eam) the C /I ma y v ary signic an tly due to co-c hannel in terferenc e. In order to tak e this in to acc oun t w e rst deriv ed the noise v alue from the C / N expression: C N = P AG LK B T (12) where P is the radiate d p o w er, B is the bandw i dth , G is the a n tenna gain at the receiv er, K is the Boltzmann constan t, T is the an tenna noise temp eratur e at the recei v er, L is the path-loss and A is the arra y f a ctor at the transmitter. He nce, w e can deriv e the noise p o w er as : N = B N 0 = K B G/T (13) F rom the BGAN standard and c ommon user t e rminal parameters, w e ha v e B = 200 KHz, G/T = 12 . 5 dB and L = 187 . 05 dB. Th us the noise p o w er i s N = − 133 dBm. Note that in eac h sim ulation the C /I remains constan t, as it only dep ends on th e p os i tion of the user terminal, whil e the C / N c hanges. As b enc hmark system w e consider one in whic h a single p olarization is considered (SISO system). The same total p o w er p er b eam P = P SI SO 2 is assumed in b oth systems. In order to ev aluate the adv an tage deriving b y using b oth p olarizations w e use the normalized throughput whic h is dened as follo w s Normalized Throughput = Throughput Dual P olarization Throughput SISO , (14) where Throughput = (1 − FER ) Rate . (15) 6 of 11 American Institute of A e ronautics and Astronautics In the rst si m ulations w e do not assu m e an y correlation b et w een p olarizations giv en that the cross-p olar in terference is kno wn to b e v ery lo w in L band. A frequency reuse factor of 4 is assumed. W e ev al u a t e the p erformance of our metho d for 2 b eams represen ting the b est and the w orst case scenarios (cen ter and edge of the b eam co v erage) assuming a user lo cation b oth at the cen ter of the b eam and at the edge . Real ist i c b eam patterns ha v e b een considered and co-c hannel in terference from b eams at the same frequency as t he reference one ha v e b een t ak en in to accoun t. W e indicate with C /I the ratio b et w een the p o w er o f the reference signal and the total in terference p o w er. The p olarization and b e am gain v alues ha v e b een tak en from. 19 P erfect c hannel state information at the recei v er (CSIR) is assumed. Figures 5 and 6 sho w the throughput of the prop osed sc heme in the mariti me c hannel normalized to the throughput of the b enc hmark (SISO) system. It can b e seen ho w, giv en a co de rate, our sc heme is able to double the rate at exp enses of incremen ting the transmit p o w er b y 3dB. The impact of the C /I on the system is remark able s i nce, when considering the b eam at the end of co v erage, the p o w er incremen t neede d to ge t t wice the SISO throughput is increased up to 4-5 dB. W e also notice ho w b earers with higher c hannel co de rate do not succeed in ac hieving 100% gain. In order to enhance the throughput f or high rate b earers Figure 5. Normalized throughput v ersus transmit p o w er for p olarization-m ultiplexin g sc heme and Soft-MMSE-SIC receiv er at cen ter of co v erage. w e ev aluated the p erformance of the SIC sc heme conside r e d in Section IV for a single p olarization setup in a m ultib eam satellite system with frequency reuse 4 using the same in terference pattern as in gure and 6 , i.e., edge of b eam at edge of co v erage. W e considered the b est case scenario, i.e., A W GN and p erfec t sym b ol alignmen t across all signals. Only the strongest in terferer (whic h has a relativ e C /I of ab out 14 dB) is tak en in to accoun t b y the dete ctor/deco der, as the others ha v e m uc h smaller p o w e r and w ould determine an increase in complexit y with limited gain in terms of BLER. The results of the sim ulations are sho wn in Fig. 7 . It can b e seen ho w the SIC giv es only a marginal gain ev en in the b est c ase scenario. This is due to the strong p o w er u n balance b et w e en the reference s i gnal and the in terferers. As a matter of facts, when the SNR is suc h that the go o d signal starts to b e de deco dable, the stronger i n terferer i s to w eak to con tribute to the deco ding and th us almost no dierence is observ e d in BLER. Other sim ulations w e carried out ( not rep orted here for a matter of space ) sh o w ed that, in case the C /I relativ e to the strong in terferer and the reference s i gnal ha v e comparable p o w er, SI C can signican tly enhance system's BLER in certain b earers. This is the case, for i ns tanc e, of systems with more aggressiv e frequency reuse factors (e.g., 2 ). Th us w e conclude that, for the considered setup, SIC do es not bring signican t impro v emen ts. Hence, in the rest of the pap er w e will not consider SIC tec hniques. In the follo wing w e presen t the results w e obtained in the ITS scenario and in a mixed LMS en vironmen t (MIX scenario). F or suc h scenario, far more c hal lenging t han the maritime one, w e used real c hannel measuremen ts obtained for ITS and MIX scenarios i n the con text of ESA MIMOSA pro ject. Note that 7 of 11 American Institute of A e ronautics and Astronautics Figure 6. Normalized throughput v ersus transmit p o w er for p olarization-m ultiplexin g sc heme and Soft-MMSE-SIC receiv er at edge of co v erage. Figure 7. BLER v ersus C / N in A W GN for the case in whic h SIC is a p plied on the reference signal and the strongest in terferer in a m ultib eam sa tell ite system with frequen cy reuse 4 . The six strongest in terferers ha v e b een considered. Three iterations h a v e b een used at the est imator. Bearer F80T1Q4B-L1 with rate 0 . 825 and QPSK mo dulation ha v e b een used for all signals. correlation eects are implicitly tak en in to accoun t in the measuremen ts. In Fig. 8 part of the measured c hannel realiz ation in ITS is sho wn. The alternation of p erio ds of mo derate and deep fading can b e observ ed. Figures 9 and 10 sho w the normalized throughput for the ITS and MIX c hann e ls. It can b e seen that a certain throughput gain with resp ect to the SISO case can b e ac hiev ed ev en for the considered c hannels if the loss in terms of C / N is tolerable. Note also that no i n terlea v er is incl u de d in the considered b earers a . The inclusion of a time in terlea v er is lik ely to enhance the p erformance of the system signican tly and allo w to exploit the full p oten tial of dual p olarization transmission. The follo wing observ ations can b e made: • Giv en a QPSK mo dulation with a determined co de rate, the use of p olarization m ultiplexing join tly with a Soft-MMSE-SIC demo dulator is able to double the data rate at exp enses of increasing the transmit p o w er of 3 dB in the b est case (in terms of C /I ) and 4 − 5 dB in the w orse situations for the maritime scenario. • When considering more c hallenging scenarios (ITS,MIX), the use of dual p olarization do es not pro vide a the BGAN standard includes the use of a time in terlea v er with 80 msec depth for some higher order b earers 8 of 11 American Institute of A e ronautics and Astronautics Figure 8. ITS c hannel realization from real measuremen ts. h 11 and h 22 comp onen ts are sho wn. Figure 9. Normalized throughput v ersus transmit p o w er for p olarization-m ultiplexin g sc heme and Soft-MMSE-SIC receiv er at cen ter of co v erage for ITS c hannel. the exp ected gain. This is mainly due to t he absence of a time in terlea v er. • The fact that the results ha v e b een obtained with a lo w complexit y demo dulator mak e s the appl ication of the considered tec hniques app ealing from a practical p ersp ectiv e. F rom a system lev el p ersp ectiv e it is clear that a rate increase can b e obtaine d also b y considering more ecien t MODCODS (i.e., higher c hannel co de rates and higher order mo dulations). W e ev aluated this n umerically and concluded that the use of a dual p olarization is the b est option for the considered setup since the increase in transmit p o w e r to obtain t wice th e throughput as in the SISO case is the less dem and i ng one for a certain lev el of BLER. T o illustrate this, w e compare dieren t w a ys to double the throughput. Suc h increase ma y b e ac hiev ed using dual p olarization, increasing the co derate and increasing the mo dulation order. In 11 these tec hniques are normalized b y the baseline scenario. The most remark able asp ec t is the fact that the spatial m ultiplexing with dual p olarization is 1 dB p o w erless to ac hiev e the s a me rate. 9 of 11 American Institute of A e ronautics and Astronautics Figure 10. Normalized throughp ut v ersus transmit p o w er for p olarization-m ultiplexing sc heme and Soft-MMSE-SIC receiv er at cen ter of co v erage for MIX c hannel. Figure 11. Doubling thr o u ghput tec hniqu es comparison. VI. Conclusions and future w ork W e presen ted the results of our study on the application of p olarization-time co des and soft in terference cancellation i n m ultib eam satellite systems. W e adopt e d the BGAN standard and used reali s ti c c hannel mo dels and in terference patt e rn s. As future w ork w e plan to extend the sim ulations to higher order mo dulations, in v estigate the use of single user preco ding as describ e d in the curren t high data rate terrestrial standards. As suc h tec hniques require feedbac k from the receiv er the impact of system round-trip dela y will b e also studied. The use of the Soft-MMSE-SIC demo dulator m us t b e in v estigated also for the mobile broadcast standards where the use of long in terlea v ers migh t increase the p erformance of the system y et main taining a lo w complexit y receiv er. Another p ossible researc h line is to study systems with more aggressiv e fr e q ue ncy reuse factors using join t SIC and MIMO tec hniques, that are l ik ely to pro vide go o d results in terms of t hroughput and system a v ailabilit y . 10 of 11 American Institute of A e ronautics and Astronautics A c kno wledgemen ts The presen t w ork has b een carried o u t under the AR TES 1 programme founded b y the Europ ean Spac e Agency . The view expressed herein can in no w a y b e tak en to reect t he ocial opinion of the Europ ean Space Agency . References 1 G. Maral and M. Bousquet. Satel lite Communic ations Systems - systems, te chniques and te chnolo gies . John Wiley & Sons Ltd, 2009. 2 J.G. Andrews. In terfere nce canc el latio n for cellular systems: a con temp orary o v erview. IEEE Wir eless Commun ic ations , 12(2):1929, Apr. 2005. 3 P . Arap o glou, P . Burzigotti, M. Bertinelli, A. Bolea Alamanac, and R. De G a udenzi. T o MIMO or no t to MIMO in mobile satellite broadcasting systems. IEEE T r ansactions on Wir eless Communic ations , 10(9):28072811, 2011. 4 A. P erez-Neira, C. Ibars, J. S erra, A. del Coso, J. Gomez , and M. Caus. Mimo applicabilit y to satellite net w or ks. In International W orkshop on Signal Pr o c essing for Sp ac e Communic ations , pages 19, 2008. 5 P . Arap oglou, K. Liolis, M. Bertinelli, A. P anagop oulos, P . Cottis, and R . De Gaudenzi. Mimo o v er satellite: A review. IEEE Communic ations Surveys T utorials , 13(1):2751, 2011. 6 Europ ean Space A g ency . Next generation w a v eform for impro v ed sp ectral eciency . h ttp://telecom.esa.in t/telecom/www/o b ject/index.cfm?fob jectid=32272, 2013. 7 ETSI. Etsi ts 102 744-2-1, satellite comp onen t of um ts (s-um ts); family sl satell ite radio in ter face; part 2: Ph ysical la y er sp ecications; sub-part 1: Ph ysical la y er in terface. T ec hnical rep ort, Europ ean T elecomm unications Standards Institute, 2013. 8 S. Alamouti. A simple transmit div ersit y tec hnique for wireless comm unications. Sele cte d A r e as in Communic ations, IEEE Journal on , 16(8):14511458, 1998. 9 P .W. W olniansky , G.J. F osc hini, G.D. Golden, an d R. V alen zuela. V-blast: an a rc hitecture for realizing v er y high data rates o v er the ric h-scat tering wireless c hannel. In Signals, Systems, and Ele ctr onics, 1998. ISSSE 98. 1998 URSI International Symp osium on , pages 295300, 1998. 10 J. C Belore, G. Rek a y a, a nd E. Viterb o. T h e golden co de: a 2 times;2 full-rate space-time co de with n on v anishing determinan ts. Information The ory, IEEE T r ansactions on , 51(4):14321436, 2005. 11 S.H. M uller-W einfurtner. Co d ing approac hes for m ultiple an tenna tran s mission in fast fading and ofdm. Signal Pr o c essing, IEEE T r ansactions on , 50(10):24422450, 2002 . 12 M. Sellath urai, P . Guinand, and J. Lo dge. Space-time c o ding in mob i le satelli te comm unications u si n g dual-p olarized c hannels. V ehicular T e chnolo gy, IEEE T r ansactions on , 55(1):188199, 2006. 13 P . F ertl, J. Jalden, and G. Matz. P er f o rmance assessmen t of MIMO-BICM demo dulators based on m u tual information. Signal Pr o c essing, IEEE T r ansactions on , 60(3):13661382, 2012. 14 I.B. Collings, M. R G Butler, an d M.R . McKa y . Lo w complexit y receiv er design for MIMO bit-in terlea v ed co ded mo du- lation. In Spr e ad Sp e ctrum T e chniques and Applic ations, 2004 IEEE Eighth International Symp osium on , pages 1216, 2004. 15 D. Seethaler, G. Matz, and F. Hla w atsc h. An ecien t mmse-based demo dulator for mimo bit-in terlea v ed co ded mo dula- tion. In Glob al T ele c ommunic ations Confer enc e, 2004. GLOBECOM '04. IEEE , v olume 4, page s 24 552459 V ol.4, 2004. 16 S. V e rdu. Multi User Dete ction . Cam bridge Univ ersit y Press, 1998. 17 X. W ang and H. V. P o or. Iterativ e (turb o) soft in terference cancellation and deco ding for co d ed CDMA. IEEE T r ans- actions on Communic ations , 47(7):10461061, 1999. 18 G. Cola v olp e, D. F er tonani, and A. Piemon tese. SISO detection o v er linear c hannels with linear complexit y in the n um b er of in ter f e rers. IEEE Journal of Sele cte d T opics in Signal Pr o c essing , 5(8):14751485, 2011. 19 ESA. T ec hnical annex b: T ec hnologies, mo dels and requiremen ts for the w or k of r g2 on mss. T ec hnical rep ort, Eur op ean Space Agency , 2013. 11 of 11 American Institute of A e ronautics and Astronautics