Deliverable 3

7.3Advanced Modulation and MIMO Schemes

7.3.1Distributed Coded MIMO Application in Terrestrial TV Broadcasting

To increase area coverage, single frequency networks (SFN) are commonly used for terrestrial TV broadcasting. SFNs are based on the simple addition of lower power transmitters at various sites throughout the coverage area. In an SFN, several transmitters transmit at the same moment the same signal on the same frequency. SFN can provide lower transmitted powers, increased bit rates and better performance. A 3-dimensional (3D) space-time-space block code (STSBC) [109] for broadcasting systems with an SFN deployment supporting mobile and portable reception is presented in this section. The use of one additional space dimension is due to the SFN deployment. The STSBC consists of 2 layers: one layer corresponding to an inter-cell ST coding, the second corresponding to an intra-cell ST coding.
More precisely, we consider a distributed MIMO scheme using (2×N_t) transmit antennas (Tx) and N_r receive antennas (Rx), which represents a SFN deployment with 2 geographically separated transmit sites. Each site implements N_t transmission antennas. The transmission could therefore be seen as a double layer scheme in the space domain. The first layer is seen between the 2 sites separated by D km (distributed MIMO scheme). The second layer is seen between the antennas separated by d m within one site. For the first layer, an STBC encoding scheme is applied between the 2 signals transmitted by each site. In the second layer, we use a second STBC encoder for each subset of N_t signals transmitted from the same site. For the first layer, the STBC encoder takes L sets of Q data complex symbols each (s₁,…,s_Q) and transforms them into a 2×U output matrix according to the STBC scheme. In the second layer (the second step), the encoder transforms each component of the first layer matrix into N_t ×T output matrix according to the second layer STBC scheme. The number of rows of the encoding matrix in the first layer is equal to two since, the STBC scheme is applied between the signals of two different sites. The output signal of each site is fed to N_t OFDM modulators, each using N sub-carriers. The reader could construct a double layer Alamouti code, for example, by considering 2 sets of 2 symbols each and then, by applying Alamouti encoding between the 2 symbols’ sets and another Alamouti encoding between the signals in each site. More generally, the double layer encoding matrix is described by:

(7.3)

In (7.3 ), the superscript indicates the layer, is a function of the input complex symbols sq and depends on the STBC encoder scheme. The time dimension of the resulting 3D code is equal to U×T and the resulting coding rate is. In order to have a fair analysis and comparison between different STBC codes, the signal power at the output of the ST encoder at each site is normalized by 2×N_t.
We restrict our study to N_t =2 transmit antennas by site and N_r =2 receive antennas. We construct the first layer with the Alamouti scheme, since it is the most resistant for the case of unequal received powers. In a complementary way, we propose to construct the second layer with the Golden code since it offers the best results in the case of equal received powers. After combination of the 2 layers, (7.3 ) yields:

(7.3)

where , , , and (.)* stands for complex conjugate.

_{Figure : Performance of 3D MIMO codes, Required Eb/N0 to obtain a BER=10-4, double layer case, DVB-T parameters,}

=4 [b/s/Hz]: Alamouti code: 64-QAM, R=1, Rc =2/3.

Other schemes: 16-QAM, R=2, Rc =1/2.

=6 [b/s/Hz]: Alamouti code: 256-QAM, R=1, Rc =3/4.

Other schemes: 64-QAM, R=2, Rc =1/2.
In the following, we will compare different STBC schemes assuming that a portable or mobile terminal receives signals from the 2 sites with unequal powers. It is a real case in SFN where the terminal receives signals from the 2 sites transmitters. We will assume that the relative power imbalance factor between the received signals from the two sites is equal to β. At the receiving side, we assume that a sub-optimal iterative receiver is used for non-orthogonal STBC schemes. The sub-optimal solution proposed here consists of an iterative receiver where the ST detector and the channel decoder exchange extrinsic information in an iterative way until the algorithm converges.
Figure shows the results in terms of required E_b/N₀ to obtain a BER equal to 10^-4 for different values of β and 3 STBC schemes i.e. the 3D code scheme, the 1-Layer Alamouti and the Golden code schemes. It can be seen that our proposed scheme presents the best performance whatever the spectral efficiency and the factor β. Indeed, it is optimized for SFN systems owing to the robustness of the Alamouti scheme to unbalanced received powers and the full rank of the Golden code. For β = -12 dB, the proposed 3D code offers a gain equal to 1.8 dB (respectively 3 dB) with respect to the Alamouti scheme for  = 4 [b/s/Hz] (resp.  = 6 [b/s/Hz]). This gain is even greater when it is compared to the Golden code. Moreover, the maximum loss of our code due to unbalanced received powers is only equal to 3 dB in terms of E_b/N₀. These results confirm that the proposed 3D code is very robust whatever the spectral efficiency and the imbalance factor β. Eventually, we should note that the factor β could be related to the channel impulse response delay and to the power path loss. Then, it can be used to adjust synchronisation problems.

7.3.2Distributed Coded MIMO Application in Hybrid Terrestrial / Satellite TV Broadcasting

_{Figure : Hybrid terrestrial/satellite transmission scheme for DVB-NGH.}

Besides terrestrial broadcasting, the DVB-NGH system can operate in combination with a satellite component to enhance the coverage and the quality of the service in a large area. As shown in Figure , the satellites as well as the terrestrial sites are used to transmit the same program to the TV receivers. The transmission can be arranged as either SFN or MFN. In the following presentation, we will focus on the SFN deployment where the philosophy of the 3D MIMO coding presented in the previous section can also be adopted [110].
More precisely, the satellites and terrestrial sites can be viewed as two different “cells”. Each cell consists of purely satellite or terrestrial transmitters. Therefore, a space-time-space block code can be formed based on a two level construction of ST coding: inter-cell (i.e. satellite-terrestrial signal) coding and intra-cell (i.e. intra-satellite and intra-terrestrial, respectively) coding. Hence, the double layer encoding matrix can also be described by (7.3 ) given in section 7.3.1. A very important work is to find the suitable ST coding scheme for each layer.
In order to construct the first layer, i.e. the ST encoding between the terrestrial and satellite sites, we consider one antenna by each site. In other words, one terrestrial site with single transmit antenna cooperates with one satellite. The second layer matrix X⁽²⁾ in (7.3 ) thus turns to one element. Due to the mobility, the mobile receiver is assumed to move among different locations. According to the typical land mobile satellite (LMS) channel model [106], the signal transmitted from satellite may experience moderate or deep shadowing. Therefore, the first layer ST scheme must be efficient face to shadowing.
In the sequel, we present first the results obtained with different elevation angles and spectral efficiencies using an Alamouti and Golden code scheme at the first layer. The main simulation parameters are list in Table . Two spectral efficiencies, i.e. η = 2, and 6 b/s/Hz, are evaluated in the simulation. The corresponding modulation and channel coding to reach these spectral efficiencies are list in Table . COST 207 TU-6 channel model [107] is used for terrestrial link while the narrow-band LMS is used for the satellite link. The received signal is assumed to follow Loo distribution [108]. The value of different parameters in the Loo distribution is given in Table .
Figure shows the required E_b/N₀ to obtain a BER equal to for a spectral efficiency of , and 6 b/s/Hz and two elevation angles. The results depicted in this figure show that for deep shadowing levels (), the Alamouti scheme presents almost better results. The Golden code presents better results only for low shadowing levels () and high spectral efficiency ( b/s/Hz).
_{Table : System parameters used in the simulations}

Parameter	Value
Sampling frequency	9.14 MHz
FFT size	256
Guard interval	1/4
Modulation	QPSK, 16-QAM, 64QAM and 256QAM
Channel coding	Convolutional codes
Code rates	1/4, 1/3, 2/5, 1/2, 3/5, 2/3
Channel estimation	ideal
Spectral efficiencies	η = 2, and 6 b/s/Hz

_{Table : Modulation and channel coding for different MIMO schemes.}

Spectral efficiency	ST scheme	ST coding rate	Constellation	Channel coding rate
η = 2 b/s/Hz	Alamouti	1	16 QAM	1/2
	Golden	2	QPSK	1/2
	3D code	2	QPSK	1/2
η = 6 b/s/Hz	Alamouti	1	256 QAM	3/4
	Golden	2	64 QAM	1/2
	3D code	2	64 QAM	1/2

_{Table : Average Loo model parameters in dB for various angles.}

Elevation	Case 1: LOS			Case 2: moderate shadowing			Case 3: deep shadowing
	M		MP	M		MP	M		MP
	-0.1	0.5	-19	-8.7	3	-12	-12.1	6	-25
	-0.5	1	-15	-4.7	1.5	-19	-7	3	-20
	-0.5	1	-17	-6.5	2.5	-17	-14	2.5	-20

Considering the whole 3D MIMO code construction, we have to find ST coding scheme for the second layer of the hybrid system, that is, the ST coding for inside each “cell”. The proposed code should be robust face to the low, moderate and deep shadowing levels. In the first step, we restrict our study to two transmit antennas by cell. According to the results from Figure , we propose to construct the first layer with Alamouti scheme, since it is the most resistant for deep shadowing case. In a complementary way, we propose to construct the second layer with the Golden code since it offers the best results in the case of low shadowing levels and high spectral efficiency. After combination of the two layers, it yields the same 3D MIMO code (7.3 ) as in the terrestrial broadcasting case.

Figure shows the results in terms of required Eb/N0 to obtain a BER equal to for different values of and 3 STBC schemes including our proposed 3D MIMO scheme. The results obtained in this figure assume that the transmission from the satellite to the receiver is achieved through a LMS channel where a COST 207 TU-6 channel model is assumed for terrestrial transmission. This figure shows the superiority of our 3D MIMO scheme. Its gain compared to the Alamouti scheme is about 1 dB for η= 2b/s/Hz and it could reach 4 dB for η = 6 b/s/Hz. Moreover, its performance remains quasi unchanged when the terminal changes its elevation angle . This means that it leads to a powerful code for SFN architectures and NGH systems.

_{Figure : Required Eb/N0 to obtain a BER=10-4, single layer case.}

_{Figure : Required Eb/N0 to obtain a BER=10-4, double layer case.}