INTRODUCTIONThe mobile Internet has finally arrived with theworldwide deployment of high-speed packetaccess (HSPA) networks and broad availabilityof third-generation (3G) terminals, mobilebroadband USB sticks, and, increasingly, note-books with integrated HSPA modules. With flat-rate data tariffs, the usage of mobile Internethas skyrocketed in 2008. Third-generation tech-nology was developed more than a decade ago,and the uptake after launch was below expecta-tions in many cases. There are various reasonsfor that, including initial lack of handset avail-ability and initial technology performance belowpredictions.

The Next Generation Mobile Networks(NGMN) Alliance has set out requirements forfuture mobile networks [1], and the Third Gen-eration Partnership Program (3GPP) is address-ing them with the development of long-termevolution (LTE). Among the requirements forLTE are increased average and peak data rates,reduced latency, spectrum flexibility addressingbandwidths of up to 20 MHz, and, last but notleast, reduced cost of ownership. The targets inNGMN and LTE are set challenges to ensure asignificant performance step from HSPA to anew technology generation.

The performance of LTE meets the essentialNGMN requirements, but not the preferred

requirements in important key performance indi-cators (KPIs) like spectral efficiency and cell-edge throughput. Therefore, development ofLTE technology is continuing beyond Release 8to address operator requirements as well asthose of the International TelecommunicationsUnion (ITU) for future technologies in thenewly identified spectrum. 3GPP has initiatedthe LTE-Advanced study item and definedrequirements in [2].

The research project Enablers for AmbientServices and Systems Part C Wide Area Cov-erage (EASY-C) is developing technologies forfuture wireless systems such as LTE-Advanced.The special feature of EASY-C is that researchideas are tested in research field testbeds at thesystem level. In EASY-C, 16 partners worktogether across the value chain, including aca-demic institutions, mobile operators, networkinfrastructure, antenna, and test equipment pro-viders, terminal chipset vendors and semiconduc-tor companies, and network planning specialists.

OVERVIEW OF LTE RELEASE 8The radio interface of 3GPP LTE/SAE Release8 uses orthogonal frequency-division multipleaccess (OFDMA) with cyclic prefix in the down-link and single-carrier frequency-division multi-ple access (SC-FDMA) with cyclic prefix in theuplink. The physical layer of LTE is defined in abandwidth agnostic way and supports varioussystem bandwidths up to 20 MHz. Radioresources are subdivided into physical resourceblocks (PRBs) consisting of 12 subcarriers with15 kHz spacing and a time duration of 1 ms.PRBs are dynamically allocated to users in orderto realize multi-user diversity gain in both timeand frequency domains, leveraging adaptivemodulation and coding (AMC) with hybrid auto-matic repeat request (HARQ).

To meet the performance requirements [3],LTE Release 8 relies on multi-antenna-basedmultiple-input multiple-output (MIMO) trans-mission and reception techniques, with 2 2MIMO as the baseline for downlink and 1 2

IEEE Communications Magazine February 200992 0163-6804/09/$25.00 2009 IEEE

ABSTRACT

The 3GPP LTE standard is stable now in its firstrelease (Release 8), and the question is howgood its performance is in real-world scenarios.LTE is also a good base for further innovations,but it must be proven that they offer perfor-mance advantages for the price of their complex-ity. This article evaluates the performance ofLTE Release 8 as a baseline and advanced con-cepts currently in discussion such as cooperativeMIMO based on system-level simulations, andmeasurements in the laboratory and a multisitefield testbed within the EASY-C project.

LTE 3GPP RELEASE 8

Ralf Irmer, Vodafone

Hans-Peter Mayer, Andreas Weber, Volker Braun, Michael Schmidt, Michael Ohm, and Norbert Ahr,

Alcatel-Lucent Bell Labs

Andr Zoch, Signalion GmbH

Carsten Jandura, Patrick Marsch, and Gerhard Fettweis, Technische Universitt Dresden

Multisite Field Trial for LTE and Advanced Concepts

IRMER LAYOUT 1/19/09 3:41 PM Page 92

IEEE Communications Magazine February 2009 93

MIMO for uplink. However, higher order anten-na configurations are supported. In the downlinkclosed-loop MIMO with code-book-based linearprecoding can be applied, which allows for spa-tial multiplexing with dual code-word transmis-sion on up to four transmission layers with fastrank adaptation. Additionally, an Alamouti-typetransmit-diversity technique called space-fre-quency block coding (SFBC) is supported. In theuplink multi-user (virtual) MIMO is used forcapacity enhancement, in which pairs of spatiallynear-orthogonal users may transmit concurrentlyon the same physical resource blocks.

EVALUATION OFNEXT-GENERATION NETWORKS

As mentioned previously, the performance of 3Gnetwork equipment and terminals was not veri-fied to the full extent when 3G was launched inearly deployments. This is one lesson learned;therefore, NGMN [1] requested performanceevaluation and field trials in parallel to standardsdevelopment. NGMN and 3GPP have initiatedthe LTE/SAE Trial Initiative (LSTI), which con-ducts trial activities and facilitates interoperabili-ty tests of LTE equipment.

Provided the metrics are meaningful and themethodology reflects realistic networks, simula-tions are a good way to compare different con-cepts and predict absolute values of networkperformance. The NGMN performance evaluationmethodology [4] is well established and allowscomparison of different standards. However, thereare still a lot of effects that are hard to foresee ormodel realistically in simulations; therefore, fieldtrials are essential to assess the performance. Also,field trials are a good proof of concept for innova-tive system-level concepts with lots of interdepen-dencies such as advanced concepts addressinginterference. Field trials also allow the calibrationof simulations and allow research and develop-ment to be focused on tackling the right issues.

Cellular networks cannot be characterizedwell by single links. Interference, resource allo-cation, and propagation environment all impactsystem performance. To capture all effects, asufficiently high number of interferers must bepresent, and multiple sectors and sites have tobe involved.

Simulations and field trials focus mainly ontechnical KPIs such as throughput and latency.However, it has to be kept in mind that the ulti-mate criterion for the mobile Internet is userexperience. This is hard to define, depends onparticular applications, and changes over time and is beyond the scope of this article.

EASY-C: A FIELD TESTBED INDRESDEN

FIELD TESTBED AREA ANDMEASUREMENT SCENARIO

Two testbeds have been built and operated with-in the above mentioned research project EASY-C. In this article we concentrate on a physicallayer focused testbed in downtown Dresden,

Germany, using existing 2G/3G network sites ofoperators Vodafone and T-Mobile. Both opera-tors are also involved in the trials. An additionaltestbed focused on applications enabled throughLTE and advanced concepts is being set up inBerlin.

The chosen testbed location in downtownDresden covers various propagation conditions,which are of special interest for evaluation offourth-generation (4G) systems with MIMOlinks and interference conditions typical in fre-quency reuse one networks like LTE, and for thedevelopment of advanced algorithms such ascooperative MIMO: A representative area of a medium-sized

European city Hills in the south causing signal reflections A river through the city causing superrefrac-

tions and tropospheric refraction Urban areas with multistory buildings, lead-

ing to shadowing effects An average intersite distance of 500 m

The testbed is being built in three phases. Inthe first phase, one site with three cells startedoperating in April 2008. This central site is locat-ed near Dresdens main railway station, as shownin Fig. 1. The antenna height is 55 m. The sec-ond phase will cover a tier of sites around thiscentral site and consist of six sites with a total of18 cells. In the final stage the testbed will com-prise 10 sites with a total of 25 cells. Additionalinterferers will surround outer cells in order toemulate the interference intensity and distribu-tion of a network with three tiers of sites. Such arather extensive setup is necessary to capture alleffects of a real-world deployment.

n Figure 1. Field test area in Dresden.

Vodafone sites

Phase 1

T-Mobile sites

Phase 2 Phase 3

IRMER LAYOUT 1/19/09 3:41 PM Page 93

IEEE Communications Magazine February 200994

At these locations new base station antennas,feeders, and microwave link equipment areinstalled.

With this three-tier network, realistic scenar-ios can be set up to investigate LTE andadvanced algorithms beyond LTE Release 8.

Furthermore, the baseline trial setup consistsof a testbed platform of base stations and mobileequipment provided by the project partner Sig-nalion. Other project partners equipment (i.e.,base stations, mobile, and chip prototypes) willbe inserted into the testbed for various test cases.

This infrastructure enables well defined andreproducible interference scenarios in bothuplink and downlink.

FIELD TESTBED EQUIPMENT

Figure 2 shows a test base station with a base-band unit, radio frequency (RF) hardware(including duplex filters and power amplifiers),antenna columns, and microwave backhaul units.LTE does not require GPS controlled referenceclocks for synchronization, but they are includedin this trial to investigate advanced multicellalgorithms. The sensitivity of these algorithms tosynchronization erro