Multiresolution MBMS transmissions for MIMO UTRA LTE systems

Hierarchical constellations constitute a simple technique for achieving multiresolution and, therefore, are appealing for MBMS (Multimedia Broadcast and Multicast Service). In this paper we consider the use of M-QAM hierarchical constellations (Quadrature Amplitude Modulation) combined with MIMO (Multiple Input Multiple Output) for the transmission of multicast and broadcast services in UTRA (Universal Mobile Telecommunications System Terrestrial Radio Access) Long Term Evolution (LTE) systems based on Orthogonal Frequency Division Multiplexing (OFDM). Due to the demanding channel estimation requirements and the high sensitivity to interference resulting from the usage of several antennas and hierarchical constellations, an enhanced receiver based on the turbo concept is employed and its performance is evaluated.


INTRODUCTION
It is widely recognized that OFDM modulations [1] are suitable for broadband wireless systems.For this reason they were selected for several digital broadcast systems and wireless networks [2] and are also being considered for UTRA LTE [3].Regarding UTRA LTE, special attention is being devoted to the support of MBMS which has already been standardized in 3GPP UTRAN (UMTS Terrestrial Radio Access Network) Release-6 [4] and 7 [5].The goal is to enable an efficient support of downlink streaming (from the base station to the mobile terminal) and download-and-play type services to large groups of users.From the radio perspective, MBMS includes point-to-point (PtP) and point-to-multipoint (PtM) modes.Regarding the PtM mode it seems attractive to employ hierarchical modulations since it is a simple and flexible enhancement technique that can increase the transmission efficiency, due to its ability to provide unequal error protection to different bits and thus provide multiresolution into a cell.By having several classes of bits with different error protection associated and to which different streams of information are mapped, a given user can attempt to demodulate only the more protected bits or also the bits that carry the additional information, depending on the propagation conditions.This type of approach is possible whenever the information can be scalable like the cases of coded voice or video signals, as studied in [6] [7].For this reason hierarchical 16-QAM and 64-QAM constellations have already been incorporated into DVB-T (Digital Video Broadcasting -Terrestrial) standards [8].
MIMO schemes have emerged as one of the most promising methods for capacity increase in a communication system [9][10] and are being considered for UTRA LTE [3].In MIMO systems with coherent detection the channel estimation plays a crucial role since the performance of the spatial signal processing in the receiver depends on the accuracy of the channel estimates.Furthermore QAM constellations can be severely affected due to inaccurate channel estimates.
In this paper we consider the use of QAM hierarchical constellations in a UTRA LTE OFDM based system employing multiple transmitting and receiving antennas with the aim of supporting broadcast and multicast services.To deal with the high sensitivity to channel estimation errors we employ an iterative receiver capable of performing joint MIMO detection and channel estimation.This receiver is based on the approach proposed in [11] for WCDMA systems.It can apply different MIMO equalization techniques during the iterative process and obtain refined channel estimates by considering the data symbols as extra pilots, as proposed in [12].
The paper is organized as follows.First Section II introduces hierarchical constellations and defines the model of the MIMO-OFDM system considered in this study.In Section III the proposed iterative receiver structure and respective channel estimation process are described.Section IV presents some performance results obtained with the proposed scheme while the conclusions are given on Section V.

A. M-QAM Hierarchical Signal Constellations
In hierarchical constellations there are two or more classes of bits with different error protection and to which different streams of information can be mapped.By using non uniformly spaced signal points (where the distances along the I or Q axis between adjacent symbols are different) it is possible to modify the different error protection levels.As an example, a nonuniform 16-QAM constellation can be constructed from a main QPSK constellation where each symbol is in fact another QPSK constellation, as shown in Figure 1.The basic idea is that the constellation can be viewed as a 16-QAM constellation if the channel conditions are good enough or as a QPSK constellation otherwise.In the latter situation, the received bit rate is reduced to half.These constellations can be characterized by the parameter k 1 =D 1 /D 2 (0<k 1 ≤0.5), as shown in Figure 1.If k 1 =0.5, the resulting constellation corresponds to a uniform 16-QAM.This approach can be naturally extended to any QAM constellation size M where the number of possible classes of bits with different error protection is

B. Transmitted Signals
In Figure 2 we show a transmitter chain that incorporates QAM hierarchical constellations into a UTRA LTE based MIMO-OFDM transmission.In the proposed scheme, there are 2 1/ 2 log M ⋅ parallel chains for the different input bit streams that will have unequal error protection.For 16-QAM we can use two parallel chains while for 64-QAM we can use three chains.Each stream is encoded, interleaved and mapped into the constellation symbols in the modulation mappers according to the importance attributed to the chain.Pilot symbols are inserted into the modulated data sequence which is then converted to the time domain using an IDFT (Inverse Discrete Fourier Transform).The resulting stream is then split into several smaller streams which are transmitted simultaneously by M tx transmitting antennas.Note that in the proposed scheme the coding is not performed independently for each different antenna.Instead each data sequence is encoded and divided equally among the transmitting antennas by the Serial to Parallel block.The objective is to try to obtain some diversity for the same encoded block.In this paper we consider the frame structure of Figure 3 for a MIMO-OFDM system with N carriers.According to this structure the pilot symbols are multiplexed with the data symbols using a spacing of with T s denoting the symbol duration, N G the number of samples at the cyclic prefix ( , , ) and h T (t) the adopted pulse shaping filter.

A. Receiver Structure
To achieve reliable channel estimation and data detection we employ a receiver capable of jointly performing these tasks through iterative processing.The structure of the iterative receiver is shown in Figure 4 where N rx receiving antennas are employed.According to the figure, the signal, which is considered to be sampled and with the cyclic prefix removed, is converted to the frequency domain after an appropriate size-N DFT operation.If the cyclic prefix is longer than the overall channel impulse response the resulting sequence received in antenna n can be expressed as The sequences of samples (2) enter the MIMO equalizer (Spatial Demultiplexer block) which separates the simultaneous transmitted streams.This can be accomplished with an MMSE (Minimum Mean Squared Error) equalizer [13], a ZF (Zero Forcing) equalizer [13], a Maximum Likelihood Soft Output criterion (MLSO) or an interference canceller (IC) [11].It is possible to perform some of the receiver iterations using one spatial demultiplexing technique, like the MMSE, and the others using a different one, like the IC, as was studied in [11].In any case, after MIMO equalization the demultiplexed symbol sequences are serialized and pass through the demodulator, de-interleaver and channel decoder blocks.This channel decoder has two outputs.One is the estimated information sequence and the other is the sequence of log-likelihood ratio (LLR) estimates of the code symbols.These LLRs go through the Decision Device, which outputs either soft-decision or hard decision estimates of the code symbols, and enter the Transmitted Signal Rebuilder which performs the same operations of the transmitter (interleaving, modulation, conversion of serial to parallel streams).The reconstructed symbol sequences are then used for a refinement of the channel estimates and also for possible improvement of the spatial demultiplexing task (in case of employing an IC) for the subsequent iteration.
The possible MIMO equalization techniques are now going to be briefly described.Using matrix notation the MMSE estimates of the transmitted symbols in subcarrier k and OFDM block l is given by ( ) where , ˆk l S is the M tx ×1 estimated transmitted signal vector with one different transmit antenna in each position, , ˆk l H is the N rx ×M tx channel matrix estimate with each column representing a different transmit antenna and each line representing a different receive antenna, , k l R is the N rx ×1 received signal vector with one different receive antenna in each position and σ 2 is the noise variance.The ZF estimate can be simply obtained by setting σ to 0 in (3).
In the MLSO criterion we use the following estimate for each symbol , , , ˆm where s i corresponds to a constellation symbol from the modulation alphabet Λ, [ ] E ⋅ is the expected value, ( ) represents a probability and ( ) p ⋅ a probability density function (PDF).Considering equiprobable symbols ( ) where M is the constellation size.The PDF values required in (4) can be computed as , ˆ( ,:) 1 , e x p 2 2 where interf , k l

S
is a (M tx -1)×1 vector representing a possible combination of symbols transmitted simultaneously by all antennas except antenna m, s is a M tx ×1 vector composed by An IC can also be used inside the Spatial Demultiplexer block, but usually is only recommendable after the first receiver iteration [11].In this case, in iteration q, for each transmit antenna m and receive antenna n, the IC subtracts the interference caused by all the other antennas.This can be represented as where ( )

B. Channel Estimation
To obtain the frequency channel response estimates for each transmitting/receiving antenna pair the receiver applies the following steps in each iteration: (1) The channel estimate between transmit antenna m and receive antenna n for each pilot symbol position, is simply computed as where , , corresponds to a pilot symbol transmitted in the k th subcarrier of the l th OFDM block using antenna m.Obviously not all indexes k an l will correspond to a pilot symbol since 1 (2) Channel estimates for the same subcarrier k, transmit antenna m and receive antenna n but in time domain positions (index l) that do not carry a pilot symbol can be obtained through interpolation using a finite impulse response (FIR) filter with length W as follows: where t is the OFDM block index relative to the last one carrying a pilot (which is block with index l) and j t h are the interpolation coefficients of the estimation filter which depend on the channel estimation algorithm employed.There are several proposed algorithms in the literature like the optimal Wiener filter interpolator [14] or the low pass sinc interpolator [15].
(3) After the first iteration the data estimates can also be used as pilots for channel estimation refinement.

IV. NUMERICAL RESULTS
To study the behaviour of the proposed MIMO-OFDM scheme and respective iterative receiver, several simulations were performed for a 16-QAM (k 1 =0.4) hierarchical constellation.Two classes of bits with different error protection were used.
Each individual information stream was encoded with a block size chosen so that the final encoded and modulated stream fitted a sub-frame composed of 7 OFDM blocks (corresponding to a 0.5ms duration).All the parameters used for these simulations were based on UTRA LTE 3GPP documents [16] and [17], for a 10MHz bandwidth.Table 1 shows the respective parameters.The channel impulse response is based on Vehicular A environment [19] with Rayleigh fading assumed for the different paths.A velocity of 30 km/h was employed unless otherwise stated.The channel encoders were rate-1/2 turbo codes based on two identical recursive convolutional codes characterized by G(D) = [1 (1+D 2 +D 3 )/(1+D+D 3 )] [18].A random interleaver was used within the turbo encoders.Most of the BER (Bit Error Rate) results presented next will be shown as a function of E S /N 0 , where E S is the average symbol energy and N 0 is the single sided noise power spectral density.
For channel estimation purposes, pilot symbols were distributed using a spacing of 6 and 4 T N ∆ = or 7 (the two possible configurations proposed in [16]) and a sinc filter interpolation with length W=2 was used at the receiver.In the graphs legends, MPB designates most protected bits, IPB means intermediate protected bits and LPB corresponds to least protected bits.
Figure 5 compares the performance of the different receiver methods of Table 2 for a MIMO 2x2 transmission employing a 16-QAM (with k 1 =0.4) hierarchical constellation.It is visible that, although the receiver with the MMSE equalizer alone performs worse than when using the MLSO equalizer, the performance can be substantially improved and achieve lower BLERs than with the MLSO when applying also an IC in the last receiver iterations.For the remainder of the paper the receiver configuration considered will be method 2 (MMSE+IC).
Figure 6 shows the behaviour of the receiver for different velocities.According to the results, the performance is almost insensitive to velocity until 120 km/h, being visible only a small degradation in the performance of the LPB.For higher velocities the performance quickly degrades for the LPB but does not change significantly for the MPB even at 300 km/h. Figure 7 compares the performance of a MIMO 2x2 transmission employing a 16-QAM (with k 1 =0.4) hierarchical constellation with the two possible pilot spacings in the time domain.It is visible that both cases have similar performances (close to the perfect estimation curves), which means that it is possible to adopt the larger pilot spacing and therefore increase the transmission efficiency if one or two antennas (we verified that the conclusion is also true even for four transmit antennas).It is important to remember however that reducing the number of pilot symbols will sacrifice the system robustness for higher velocities as will be shown further ahead.

V. CONCLUSIONS
In this paper we have studied the use of QAM hierarchical constellations with the aim of supporting multicast and broadcast transmissions in a MIMO-OFDM system similar to the one being considered for UTRA LTE.
It was verified through simulations that the iterative receivers schemes studied are able to achieve good performances for all the bit streams, including those with lower error protection levels even for very high velocities.Therefore the proposed transmitter/receiver scheme can provide unequal error protection which is adequate for supporting MBMS transmissions in UTRA LTE.It was also observed that if a maximum of four transmit antennas are being used with 16-QAM hierarchical modulations, the option with only half of the pilots symbols proposed for UTRA LTE ( 7 ) is adequate, unless a higher robustness is desired for high velocities for all information streams.

S
the frequency domain.To avoid interference between pilots of different transmitting antennas, FDM (Frequency Division Multiplexing) is employed for the pilots, which means that pilot symbols cannot be transmitted over the same subcarrier in different antennas.Data symbols are not transmitted on subcarriers reserved for pilots in any antenna, therefore, the minimum allowed spacing in the frequency domain is () is the symbol transmitted by the k th subcarrier of the l th OFDM block using antenna m.The transmitted OFDM signals are then expressed as

Figure 3 .
Figure 3. Frame structure for a OFDM transmission (P -pilot symbol, Ddata symbol, T s -symbol duration).

H
denoting the overall channel frequency response between transmit antenna m and receiving antenna n for the kth frequency of the lth time block and , m k l N denoting the corresponding channel noise.
symbols estimates of the previous iteration for transmit antenna m, subcarrier k and OFDM block l.