Impact of LTE Precoding for Fixed and AdaptiveRank Transmission in Moving Relay Node System

Ayotunde O. Laiyemo, Pekka Pirinen,and Matti Latva-aho

Centre for Wireless CommunicationsUniversity of Oulu, Finland

P.O. Box 4500, FI-90014 University of OuluEmail: forename.surname@ee.oulu.fi

Jaakko Vihriala and Vinh Van PhanNokia Siemens Networks

Kaapelitie 4, FI-90651 OuluEmail: forename.surname@nsn.com

Abstract—Improved capacity and reliability of wireless net-works can be achieved by users onboard in a high speed train withthe use of moving relay nodes (MRN) cooperating with each other.Further improvement in performance is achieved with codebookbased precoders adopted by third generation partnership project(3GPP) for spatial multiplexed transmission mode. Open loop andclosed loop precoded spatial multiplexing for fixed and adaptiverank transmissions were simulated and analyzed on the backhaullink of a cooperative MRN system in a high speed train. Toobtain improved performance, channel information in the formof precoder matrix indicator (PMI) and rank indicator (RI) arerequired for the closed loop operation while the RI is only neededfor the open loop operation.

The results show that throughput gain on the backhaul link ofthe cooperative MRN system can be achieved with the precodedtransmission when compared to unprecoded spatial multiplexingtransmission. Further improvement in throughput was achievedwhen the transmission rank changed dynamically according tothe channel condition.

I. INTRODUCTION

The long term evolution (LTE) based mobile telecom-munication system achieves higher data rate and improvedspectral efficiency with the use of multiple input multipleoutput (MIMO) techniques, and variants of orthogonal fre-quency division multiplexing (OFDM) used for transmissionmitigate against inter symbol interference (ISI). Additionalfeatures, such as relaying technologies, have been consid-ered as vital elements in the fulfillment of InternationalMobile Telecommunications Advanced (IMT-Advanced) re-quirements specified by the International TelecommunicationUnion-Radiocommunication (ITU-R) [1]. Relays are low pow-ered and low cost eNodeB devices that connect the corenetwork through a donor eNodeB with wireless backhaul link.

Fixed relay nodes are deployed near the cell edge toincrease system capacity and extend cell coverage area [2] atreduced cost without using cable or fiber [3]. The increasingpace at which large high speed transportation vehicles arebeing deployed, results in additional challenges to maintaingood quality service for users onboard. Hence support formobile relays on high speed vehicles has been considered bythe 3GPP to solve challenging issues arising from speeds ashigh as 350 km/h and high penetration loss [4].

Recent studies have shown the practicability and potentialof the use of moving relays on public transport vehicles. Forexample, studies in [5] show a significant reduction in calldrops arising from group handover and reduced transmissionpower as a result of the short range link between the onboardusers and the access point. The LTE/LTE-A standard hasadopted the use of codebook based precoding in its spatialmultiplexing transmission modes in order to improve the SINRand at the same time limit the signaling overhead [6]. Toexamine the impact of codebooks on the throughput of ourcooperative MRN implementation, the closed loop precodingscheme requiring feedback from the PMI and RI, was im-plemented in scenarios where the high speed train would bestationary or at low speed at certain places such as the trainstation. At high speed, the open loop precoding scheme wasimplemented on the train scenario with RI feedback.

This paper analyzes the impact of the use of LTE precodingspatial multiplexing transmission schemes on the backhaul linkof a cooperative MRN system on a high speed train with eightcarriages (e.g., the 8-car E Shinkansen). LTE codebook basedprecoders were derived for two and four transmit antennas andimplemented on the backhaul link using the LTE-Advancedcompliant system level simulator. Earlier studies in [7] haveconsidered the backhaul link to be the capacity bottleneckof the system and, with the use of a large backhaul antennaarray, the network capacity and throughput of onboard usersimproved significantly. Hence we focus on how to extensivelyimprove the backhaul link throughput and limit the effect ofdelayed CSI feedback resulting from the high speed. We showthat the backhaul throughput of the cooperative MRN systemis further improved with the use of LTE precoders at certaininstances on a realistic cellular network.

II. SYSTEM MODEL

A. Cooperative MRN Description

The moving relay deployment we address is based ondedicated moving relays in LTE/LTE-Advanced serving thetrain users in a high speed train that consists of eight carriages,illustrated in Fig. 1, where a moving relay node is installedin each of the eight carriages of the train. The eight MRNsconnect to the donor cells in a coordinated and cooperativefashion to ensure sufficiently high data rates for the multiplebackhaul link connections. The eight MRNs considered are978-1-4799-0846-2/13/$31.00 c©2013 IEEE

2013 13th International Conference on ITS Telecommunications (ITST)

978-1-4799-0846-2/13/$31.00 ©2013 IEEE 250

connected via the crX2 interface which is assumed to be ofzero latency and can be realized using wired interface such asoptical fibre or wireless interface operating outside the radioband frequency.

MRN 1 MRN 2 MRN 8crX2

BS

X2

BS

Fig. 1. Cooperative MRN system model [7].

Each of the MRNs has multiple antennas on the exterior ofthe train and spans the length of each carriage. The donor cellsconnected to the MRNs have the knowledge that the MRNsare cooperating together. The cooperating MRNs are dividedinto groups based on their CSI information and the strongestMRN link in each group is used as the backhaul link for thatgroup, while the other MRN links in each group are dropped.The choice of the optimal precoder from the codebook for eachMRN is calculated from the current CSI.

B. Signal Description

The LTE-Advanced system level simulator consists of acentral hexagonal layout consisting of (19sites × 3sectors)cells with copies wrapped around the central layout to ensureuniform interference level across the central layout. Each ofthe eNodeBs and macro UEs/MRNs in the layout are equippedwith NT transmit and NR receive antennas, respectively. Thereceived signal vector ysc ∈ CNR for subcarrier sc for eachUE/MRN can be expressed as

ysc = HscFscixsc +∑I

HIscFI

scixIsc + nsc (1)

where Hsc ∈ CNR×NT is the channel matrix for subcarrier scof the desired signal, xsc ∈ CNT is the desired transmit signalvector per subcarrier, Fsci ∈ CNT×NL is the precoding matrixof index i chosen from the codebook for the desired channel,the superscript I identifies the interfering components andnsc ∼ CN (0, N0INR

) represents the additive noise vector. Theclosed loop and open loop spatial multiplexing transmissionschemes were implemented on the MRN backhaul link ata train speed of 1 km/h and 300 km/h, respectively. TheRI indicates the number of transmission layers for spatialmultiplexing [8] and the value which is ≤ min(NT ,NR) isdefined by the level of linear dependency among the antennasin the channel.

We considered the minimum mean square error (MMSE)receiver in our simulation with the effective channel matrixdefined as H∗sc = HscFsci . (2)

The MMSE filter decouples the transmitted data streams atthe receiver with the decision variables generated as dsc =WH

scysc and the filter weight matrix Wsc ∈ CNR×NT isobtained by minimizing the mean square error defined asE[‖dsc −WH

scysc‖2]

and is given asWsc = (H∗scH∗Hsc + Rsc)

−1H∗sc . (3)The corresponding user’s SINR for subcarrier sc at the outputof the MMSE receiver is given by

γMsc=Pt ∗ diag|WH

scH∗sc|2

diag|WHscRscWsc|

(4)

where Pt is the transmission power, Rsc is the interferenceplus noise covariance matrix and diag(.) denotes the diagonalelement of the argument.

III. IMPLEMENTATION SCENARIO

In our simulation, the fast fading multipath channel modelbased on [9] was used. The channel was modeled to berealistic with a random LOS or NLOS propagation conditionleading to random correlation scenarios. The application ofprecoding in our implementation improves system performancewith signal isolation and/or beamforming (depending on theratio between the number of transmit antennas and the numberof spatial layers), when the multipath channel is unable toprovide adequate SINR at one or more of the receive antennasby increasing or equalizing the received SINR across thereceive antennas. The performance of the precoded spatialmultiplexing scheme is directly related to the received SINRand the channel correlation properties which depend on theantenna spacing and configuration.

A. Closed Loop Precoded Operation on Cooperative MRN

The LTE codebook consists of a limited set of precodermatrices with each matrix corresponding to an index num-ber. At each of the MRN receivers in each group formed,the PMI selects the appropriate index number based on theinstantaneous channel condition and the MRN with the bestchannel quality signals its PMI back to the eNodeB with theremaining MRNs in the group becoming dormant. Then theeNodeB applies the precoder matrix corresponding to the indexnumber received for its next transmission. The LTE codebookfor four and two transmit antennas were implemented in oursimulation. The precoder matrices for four transmit antennawere generated from the householder transformation to reducecomputational complexity according to

Fn4= I4 −

2unuHn

uHn un

(5)

where I4 is the 4×4 identity matrix, the vector un is obtainedfrom 8PSK elements and the superscript H denotes conjugatecomplex transposition. The codebook indices and values canbe found in [10].

B. PMI Selection for Closed Loop Operation

The PMI can be selected based on either capacity orSINR. The precoder selection presented in this paper is basedon the capacity/mutual information selection criterion whichobtains from the codebook, the appropriate precoder matrix bycalculating the mutual information experienced by the channelfor each subcarrier using the available precoding matrices in

2013 13th International Conference on ITS Telecommunications (ITST)

251

the LTE codebook. And then, choosing the precoder matrixthat gives the maximum mutual information for each subband.The selection criterion can be expressed as

Csc(Fi) = log2 det

(INL

+εs

NLN0FHi HH

scHscFi

)(6)

where INLis the NL×NL identity matrix, εs

N0is the energy per

bit to noise power spectral density ratio and NL is the numberof transmission layers.

C. Open Loop Precoded Operation on Cooperative MRN

The open loop precoded spatial multiplexing operation issuitable for high mobility scenarios because it provides anincreased level of diversity by transmitting delayed versionsof the time domain signal simultaneously from the multipleantennas of the active MRN in each group formed. The PMIis not required, only the RI feedback is used to determine thesize of the precoding matrices that will be applied cyclicallyacross the subcarriers. A subset of the LTE codebook is usedin combination with DFT matrices and large cyclic delaydiversity matrices and expressed in this form

Fsc = Fn[isc] ∗ Dsc ∗ U (7)

where Fn[isc] is a NT × NL precoding matrix from the LTEcodebook, Dsc is a cyclic delay diversity (CDD) matrix ofsize NL × NL that changes with the subcarriers, U is a fixedDFT matrix of size NL × NL and NT is the number oftransmit antennas. The CDD Dsc and the DFT U matrix aredefined for 2, 3, and 4 transmission layers in [10]. In ourimplementation for two transmit antennas, Fn[isc] was fixedwhile for four transmit antennas, Fn[isc] varied among fourprecoding matrices obtained from the LTE codebook withrespect to a cyclic increase of the index isc according to

isc = mod(floor(sc/NL), 4) + 1, (8)

where sc is the index of the subcarrier and floor(X) roundsthe expression in brackets to the nearest integer less than orequal to X .

D. Dynamic Rank Selection

In this paper, the transmission rank was estimated at theterminal (MRN) and a single rank was applied across thewhole bandwidth for each active MRN in the cell. In ourimplementation, the RI feeds back the selected rank along-side with the channel quality indicator (CQI) and/or PMI tothe eNodeB. In the closed loop transmission, dynamic rankselection was achieved with the different set of precoders fordifferent number of antennas using (6), while for the open looptransmission, rank adaptation was achieved by maximizing theobtained capacities through the SINRs at the MMSE receiverfor different number of possible transmission layers in sucha way that the number of parallel transmission streams NL

over the channels are assumed to be decorrelated at the MMSEreceiver with the total SINR spread among the layers. Based onthe SINR values, rank adaptation was achieved by evaluatingthe channel capacity for each layer at the MMSE receiver.The total channel capacity for the number of layers used fortransmission is given as

CR(γMsc) =

R∑r=1

Nsc∑sc=1

{log2

(1 +

NMRN

RγM

rsc

)}(9)

where γMrsc is the MMSE post processed SINR for each layer

in the transmission per subcarrier obtained from (4), NMRN isthe number of receive MRN antennas, and the maximum valueof R is min (NT ,NMRN ). The adaptation process is illustratedin Fig. 2.

Select number of transmissionstreams R that maximizes ( ) for each MRN on train (RANK ADAPTATION)

MRN grouping based on CSI obtained

Select MRN with best channelquality in each group formed

CSI feed back for next transmission CSI = CQI + RI

Open Loop Precoding Transmission Select index of precoder matrixfrom all precoder set that maximizes ( ) for each MRN

(RANK ADAPTATION)

MRN grouping based on CSI obtained

Select MRN with best channelquality in each group formed

CSI feed back for next transmission CSI = CQI + PMI + RI

Closed Loop Precoding Transmission

SISO/SIMO Transmission

Train speed > 15km/h

Number of Antennas > 1

Measured Channel H

Obtain SINR

Yes

Yes

No

No

96

Fig. 2. Flow diagram of the closed and open loop rank adaptation processes.

IV. PERFORMANCE EVALUATION

The impact of applying closed loop codebook based pre-coding on the backhaul throughput of the cooperative MRN isevaluated when the train is assumed stationary.

TABLE I. SIMULATION PARAMETERS.

System bandwidth 10 MHz

Number of macro UEs 570

Number of MRNs 8 (1 in each carriage)

Propagation scenario suburban macro

MRN duplex mode half duplex FDD

eNodeB→ MRN antenna config 2×2, 2×4, 4×4, 4×8

MRN transmission scheme SU-MIMO spatial multiplexing

MRN transmission mode open loop and closed loop precoding

HARQ chase combining

Receiver type MMSE

L2S interface metric MIESM [11]

Train carriage wall attenuation 20 dB

Train speed 1 km/h and 300 km/h

The throughput for four different antenna configurationswere considered with the different possible precoding setsapplied from the LTE codebook. The application of the openloop precoding was also considered at a train speed of 300km/h and was implemented with large cyclic delay diversityalongside a subsection of the LTE codebook. The main param-eters for the simulation are given in Table I and the antennaconfigurations with the corresponding precoder set applied arelisted in Table II.

With 2 transmit antennas at the eNBs and transmission rank2, three precoding choices are available in the LTE codebook,

2013 13th International Conference on ITS Telecommunications (ITST)

252

TABLE II. ANTENNA AND LTE PRECODER CONFIGURATION.

Antenna configuration (NT × NMRN ) LTE precoder (NT × NL)

2× 2 2× 2, 2× 1

2× 4 2× 2

4× 4 4× 4, 4× 2

4× 8 4× 4, 4× 2

while for 4 transmit antennas at the eNBs, 16 precoderswere derived for each transmission layer to choose from. ThePMI selection for the closed loop operation was based onmaximizing the channel mutual information. Simulations weredone with 1000 drops with 50 channel samples in each drop.The backhaul throughputs were obtained and compared withthe unprecoded spatial multiplexing transmissions of the sameantenna configuration.

A. Closed Loop

Fig. 3 shows the cumulative distribution function (CDF)of the backhaul link throughput for 2 × 2 and 2 × 4 antennaconfigurations with 2× 2 LTE precoder set at a train velocityof 1 km/h. The throughput performance of the backhaul linkis degraded slightly with the application of the 2 × 2 LTEprecoder set.

0 10 20 30 40 50 600

0.1

0.2

0.3

0.4

0.5

0.6

0.7

0.8

0.9

1

Throughput [Mbps]

CD

F

2x4 ant config with 2x2 closed loop precoding2x4 ant config with no precoding2x2 ant config with 2x2 closed loop precoding

2x2 ant config with no precoding

Fig. 3. CDF plot of 2× 2 and 2× 4 MRN backhaul throughput with 2× 2closed loop precoding at 1 km/h.

At a speed of 1 km/h, the channel is considered to berelatively constant, i.e., the coherence time of the channel isfairly large. Hence unprecoded parallel transmissions on thetwo layers are fairly orthogonal and the application of 2 × 2LTE precoder set in which the number of transmit antennas(NT ) is equal to the number of layers (NL) (full rank) doesnot provide any gain. This is because a precoder set wherethe number of transmit antennas (NT ) is equal to the numberof layers (NL) will only provide improved signal isolationbetween the two layer transmission and the widely spacedMRN antennas already provides this when the train is relativelystationary. The small degradation results from subband PMIusage.

The simulation results for the antenna configurations of4×4 with 4×2 LTE precoder set at a train velocity of 1 km/hare shown in Fig. 4. Two transmission layers are constructedusing the precoding scheme designed for 4 transmit antennas.

0 10 20 30 40 50 60 700

0.1

0.2

0.3

0.4

0.5

0.6

0.7

0.8

0.9

1

Throughput [Mbps]

CD

F

4x4 ant config with 4x2 closed loop precoding4x4 ant config with no precoding

Fig. 4. CDF plot of 4×4 MRN backhaul throughput with 4×2 closed loopprecoding at 1 km/h.

The 50% point on the CDF curve of Fig. 4 shows an increaseof 3 Mbps in the precoded throughput as compared to theunprecoded throughput. The gain in throughput is achieved bythe 4×2 precoding which provides a combination of improvedorthogonalized layer transmission and beamforming. Since thenumber of layers (NL) is less than the number of transmitantennas (NT ), the precoding provides a mapping of NT →NL achieving beamforming and with the number of receiveantennas (NMRN ) equal to the number of transmit antennas,the 2 spatial layers constructed achieve improved isolation atthe receiver side which yields the improved performance inFig. 4. Since the MRN receive antennas are widely spaced andspan the length of the train, it results in highly uncorrelatedchannels between the cooperative MRN antennas and the eNBsand the effects of the precoders applied are minimal.

B. Open Loop

Spatial multiplexing for 4× 4 and 4× 8 antenna configu-ration with the application of 4× 4 and 4× 2 LTE precodingset are presented in Figs. 5 and 6.

0 10 20 30 40 50 60 700

0.1

0.2

0.3

0.4

0.5

0.6

0.7

0.8

0.9

1

Throughput [Mbps]

CD

F

4x4 ant config with 4x4 open loop precoding4x4 ant config with 4x2 open loop precoding4x4 ant config with no precoding

Fig. 5. CDF plot of 4× 4 MRN backhaul throughput with 4× 4 and 4× 2open loop precoding at 300 km/h.

The increase in speed causes the four spatial layers con-structed to loose a certain degree of orthogonality. In both

2013 13th International Conference on ITS Telecommunications (ITST)

253

figures, we compare the unprecoded throughputs with 4 × 4and 4× 2 precoded throughput at 300 km/h.

0 10 20 30 40 50 60 70 80 900

0.1

0.2

0.3

0.4

0.5

0.6

0.7

0.8

0.9

1

Throughput [Mbps]

CD

F

4x8 ant config with 4x4 open loop precoding

4x8 ant config with 4x2 open loop precoding4x8 ant config with no precoding

Fig. 6. CDF plot of 4× 8 MRN backhaul throughput with 4× 4 and 4× 2open loop precoding at 300 km/h.

Looking at the 60% point on the CDF curves of Figs. 5and 6, we have a 3 Mbps increase in the backhaul throughputwhen comparing the unprecoded with the 4 × 4 precodedthroughput and a 8 Mbps increase is achieved with the 4× 2LTE precoder set. The gain in throughput is as a result ofimproved isolation of the four layers for the case of 4 × 4precoding set. And for the case of 4 × 2 precoder set thegain in throughput is achieved from improved isolation of thetwo layers constructed and the beamforming achieved from themapping of four transmit antennas to two transmission layers.The improved isolation is achieved by a cyclic applicationof the precoder set on all the subcarriers, thereby achievingincreased frequency diversity which improves the resilience ofthe spatial multiplexed transmissions.

C. Rank Adaptation

Fig. 7 presents a comparison of the throughput for openloop adaptive ranking at a train speed of 300 km/h with fixedrank transmission for unprecoded, 4× 4 and 4× 2 precodingset.

0 10 20 30 40 50 60 70 800

0.1

0.2

0.3

0.4

0.5

0.6

0.7

0.8

0.9

1

Throughput [Mbps]

CD

F

4x4 ant config with 4x4 open loop precoding4x4 ant config with 4x2 open loop precoding4x4 ant config with no precoding

4x4 ant config with adaptive ranking

Fig. 7. CDF plot of 4×4 MRN backhaul throughput with open loop adaptiveranking at 300 km/h.

There is a significant increase in the adaptive rankingthroughput between the 1% point and 100% point on the CDFcurve. At the 60% point on the CDF curve, the increase inthroughput compared to the unprecoded is 14 Mbps, whilethe increases in throughput compared to 4 × 4 and 4 × 2open loop precoding implementation are 11 Mbps and 6 Mbps,respectively.

V. CONCLUSION

In this paper, we analyzed the throughput gains that canbe achieved with the implementation of closed loop and openloop precoding on a cooperative MRN system deployed ona high speed train. Adaptive transmission ranking was alsoconsidered in our implementation. Our simulation results withfull rank for closed loop operation show that throughput gaincannot be achieved at relatively low speed when compared withthe unprecoded spatial multiplexing scheme. But with less thanfull rank transmission, throughput gain is achieved. Simulationresults obtained for open loop operation show performanceimprovement in the throughput gain at high speed both forfull and less than full rank transmission. Further throughputgain is achieved with adaptive ranking.

ACKNOWLEDGMENT

This research was supported by the Finnish FundingAgency for Technology and Innovation (TEKES), NokiaSiemens Networks (NSN), Renesas Mobile Europe, Elektrobitand Anite Telecoms.

REFERENCES

[1] “Requirements related to technical performance for IMT-Advancedradio interfaces ITU-R, Rep. M.2134,” Technical Specification, 2008.

[2] K. Balachandran, J. Kang, K. Karakayali, and J. Singh, “Capacitybenefits of relays with in-band backhauling in cellular networks,” inProc. IEEE Int. Conf. Communications (ICC), 2008, pp. 3736–3742.

[3] R. Schoenen, W. Zirwas, and B. H. Walke, “Capacity and coverageanalysis of a 3GPP-LTE multihop deployment scenario,” in Proc. IEEEInt. Conf. Communications (ICC), 2008, pp. 31–36.

[4] “3rd generation partnership project; technical specification group radioaccess network; mobile relay for E-UTRA, 3GPP Rep. TR 36.836V2.0.0,” Technical Specification, 2012.

[5] R. Zhao, X. Wen, D. Su, and W. Zheng, “Call dropping probability ofnext generation wireless cellular networks with mobile relay station,” inProc. IEEE 2nd Int. Conf. Future Networks (ICFN), 2010, pp. 63–67.

[6] M. Chandrasekaran and S. Subramanian, “Performance of precodingtechniques in LTE,” in Proc. IEEE Int. Conf. Recent Trends In Infor-mation Technology (ICRTIT), 2012, pp. 367–371.

[7] S. Scott, J. Leinonen, P. Pirinen, J. Vihriala, V. Van Phan, and M. Latva-aho, “A cooperative moving relay node system deployment in a highspeed train,” 2013.

[8] J. Wang, M. Wu, and F. Zheng, “The codebook design for MIMOprecoding systems in LTE and LTE-A,” in Proc. IEEE 6th Int.Conf. Wireless Communications Networking and Mobile Computing(WiCOM), 2010, pp. 1–4.

[9] J. Meinila, P. Kyosti, T. Jamsa, and L. Hentila, “WINNER II channelmodels,” Radio Technologies and Concepts for IMT-Advanced, pp. 39–92, 2009.

[10] L. Juho, H. Jin-Kyu et al., “MIMO technologies in 3GPP LTE andLTE-advanced,” EURASIP Journal on Wireless Communications andNetworking, 2009.

[11] X. He, K. Niu, Z. He, and J. Lin, “Link layer abstraction in MIMO-OFDM system,” in Proc. IEEE Int. Workshop on Cross Layer Design(IWCLD), 2007, pp. 41–44.

2013 13th International Conference on ITS Telecommunications (ITST)

254