"From LTE-Advanced to the Future"
C qM IEEE M ommunications q qM Previous Page | Contents | Zoom in | Zoom out | Front Cover | Search Issue | Next Page MqM q Qmags THE WORLD’S NEWSSTAND® LTE-ADVANCED AND 4G WIRELESS COMMUNICATIONS From LTE-Advanced to the Future Matthew Baker, Alcatel-Lucent ABSTRACT should be avoided, and careful evaluation must continue to inform the selection of new features. LTE and LTE-Advanced have undeniably In the concluding chapter of , several provided a major step forward in mobile com- aspects are identified as being of particular munication capability, enabling mobile service importance for the future evolution of LTE- provisioning to approach for the first time that Advanced. These include increased capacity and available from fixed-line connections. However, spectral efficiency, improvement of the through- market demands typically do not evolve simply put actually experienced by the user, fairness of in discrete steps; therefore, the future evolution throughput provision, reduction of cost per bit, of LTE-Advanced will be a story of continuous and energy saving. enhancement, on one hand taking advantage of The following sections explore some key the advancing capabilities of technology, while aspects of potential enhancement in more detail. on the other aiming to keep pace with the expec- tations and needs of the end users. In this arti- cle, the likely directions of this continuing SPECTRAL EFFICIENCY enhancement are discussed, and some areas While the peak data rate (or peak spectral effi- where further technical advancement will be ciency) continues to be important from a mar- required are identified. In particular, potential keting perspective, the end-user experience is measures to enhance the efficiency of spectrum more closely associated with the degree of uni- utilization by joint multicell optimization, dynam- formity of service provision; this implies that ic adaptation of the network to traffic character- future enhancements will need to augment cell istics and load levels, and support for new edge data rates rather than simply addressing applications are highlighted. peak rates. At the level of the individual radio links, pos- INTRODUCTION sibilities are limited for further improvements to the performance in a given bandwidth. Turbo The factors that have led to Long Term Evolu- codes have raised link-level spectral efficiency in tion (LTE)-Advanced becoming the leading fading propagation environments almost to the broadband mobile communications system for Shannon limit, and the performance of single- the fourth generation (4G) are in large part asso- user multiple-input multiple-output (SU-MIMO) ciated with the close partnership between net- is largely constrained by the feasible number of work operators and equipment vendors in its uncorrelated antennas (the size of the mobile development. LTE has, on one hand, been built terminal, and factors such as wind loading and to provide support for applications for which aesthetics at the base station), the characteristics there is active consumer demand, while on the of the propagation channel, and the carrier fre- other hand incorporating advanced self-optimiza- quency. It is therefore challenging to identify tion and self-configuration functionality to facili- further incremental enhancements that can be tate practical deployment for network operators. introduced at the link level without incurring This has resulted in uniquely balanced system unreasonable complexity. specifications that are technically advanced but This means that system-level enhancements practical, cost effective, and, above all, useful. must generally be targeted considering multicell The direction of the future evolution of LTE- operation in order to realize useful improve- Advanced will be guided by the principle that ments in spectral efficiency and user experience, these balances must be maintained. Not every both overall and at the cell edge. theoretical claim of potential improvement is cost effective and practical to deploy. Factors NETWORK TOPOLOGY such as robust conformance testing and perfor- Historically, mobile networks have consisted of a mance verification must also be taken into single homogeneous “layer” of macrocells of account, and care must be taken not to fragment broadly uniform size. Early deployments of LTE the market unnecessarily, in order to ensure that have required wide-area coverage to be achieved economies of scale are not lost; such factors as in a short space of time, often to meet regulato- these imply that the introduction of multiple ry requirements, and these requirements were alternative options offering similar performance most appropriately satisfied by the homogeneous 116 0163-6804/12/$25.00 © 2012 IEEE IEEE Communications Magazine • February 2012 C qM IEEE M ommunications q qM Previous Page | Contents | Zoom in | Zoom out | Front Cover | Search Issue | Next Page MqM q Qmags THE WORLD’S NEWSSTAND® C qM IEEE M ommunications q qM Previous Page | Contents | Zoom in | Zoom out | Front Cover | Search Issue | Next Page MqM q Qmags THE WORLD’S NEWSSTAND® macrocellular approach, as long as the number but there is still a need for further study to of LTE terminals was small. design effective but practical CoMP schemes Existing techniques However, as the density of LTE-equipped suitable for application in real networks. Stan- for interference terminals increases, smaller cells are needed to dardization for CoMP in future evolutions of deliver sufficient capacity; such capacity increas- LTE-Advanced will focus on the enablers for coordination es from reducing the cell size are often referred CoMP operation, including suitable reference generally focus on to as cell-splitting gains. Since the small cells are signals, measurements, and CSI feedback, while thus introduced in a pre-existing network of the actual transmission techniques (e.g., coher- interference macrocells, the result is a heterogeneous net- ent or non-coherent joint transmission, coordi- avoidance. However, work of macro- and small cells operating at dif- nated scheduling, or coordinated beamforming) future developments ferent transmission powers and with different and reception techniques (e.g., minimum mean coverage areas. The magnitude of the cell-split- squared error [MMSE]-SIC) will generally be will tend to be based ting gains is dependent on suitable cell associa- designed and optimized in proprietary vendor- on cooperative tion/handover decisions to ensure that sufficient specific ways without explicit standardization. users are served by the small cells. The CoMP schemes deployed in practical transmission rather In such networks, intercell interference man- LTE-Advanced networks will be governed by the than interference agement becomes significantly more complex availability of high-bandwidth low-latency fiber avoidance, leading than in a homogeneous network. Some intercell optic interconnections between transmission interference coordination techniques have been points. Such links are not yet widely available, so to coordinated a part of LTE since its inception, with the possi- early CoMP schemes should not be overly multipoint operation. bility of coordinating the cell edge data transmis- dependent on their existence. Nevertheless, cen- sions of neighboring cells in the frequency tralized (or cloud) processing will clearly become domain to control the intercell impact. In LTE- an important enabler for realizing the full poten- Advanced, additional possibilities for time- tial of CoMP, and efficient schemes for the domain scheduling coordination are introduced transfer of scheduling/precoding information, as , which are particularly suited to heteroge- well as data, between transmission points will neous networks of small cells overlaid on macro- need to be identified. cells; time-division multiplexing (TDM) of the It is worth noting that centralized processing control signaling and data between the different has the potential to provide some practical bene- types of cells can help to ensure that the cell- fits in addition to those arising from centralized splitting gains can be achieved to the maximum CoMP scheduling/beamforming and joint system extent. Further optimization of interference optimization: centralized processing can also coordination is one means by which overall sys- bring possibilities to reduce the costs of network tem capacity and performance is likely to be able operation by reducing the number of secure sites to be enhanced in the future, by either standard- needed for housing power-hungry, high-value, ized or proprietary mechanisms. Such mecha- and security-critical network equipment, as well nisms are likely to increasingly exploit as facilitating network reconfiguration and centralized processing in the network architec- robustness. ture in order to jointly optimize the operation of For the uplink, the benefits of CoMP recep- clusters of cells. tion are more readily achievable [5, 6], since no additional transmissions are needed compared to COORDINATED MULTIPOINT OPERATION reception at a single point. Each reception point Existing techniques for interference coordination can use interference cancellation techniques to generally focus on interference avoidance. How- isolate the weaker signals from mobile terminals ever, future developments will tend to be based in neighboring cells and convey them to a cen- on cooperative transmission rather than interfer- tral node for combining with the versions of the ence avoidance, leading to coordinated multi- signal received at other points. point (CoMP) operation. Theoretical research has provided some opti- MULTI-USER MIMO mistic and impressive indications for the poten- Within a given cell, MU-MIMO may be tial performance of CoMP with schemes known employed to increase spectral efficiency without as Network MIMO (e.g., ). In the downlink, necessarily needing the coordination or central- Network MIMO involves extending multi-user ized processing capabilities required by CoMP MIMO (MU-MIMO) at the transmit side such schemes. that the transmissions from multiple points are Downlink transmissions in the latest releases jointly precoded for coherent combining at the of LTE utilize precoded reference signals, which receivers. The gains are dependent on the num- can facilitate advanced beamforming and space- ber of cooperating transmission points, the relia- division multiplexing techniques. Nevertheless, bility of channel state information at the there remains a significant gap between the the- transmitter (CSIT) (in the case of coherent oretical promises of MU-MIMO based on dirty CoMP, full Network MIMO requires not just paper coding (DPC) techniques  and what is CSIT of a single link but CSIT for the joint achieved in practice. DPC itself will not be channel matrix for all cooperating transmission implemented in the foreseeable future, due to its points), and the bandwidth and latency of the complexity, its sensitivity to practical non-ideali- internode communication links. ties, and the unrealistic level of CSIT it requires; Practical CSI feedback considerations have a it should be seen simply as an interesting refer- significant impact on CoMP performance . ence point to indicate the upper bound of theo- Much investigative work has already been under- retical performance. taken in the Third Generation Partnership Pro- A number of reasons may be identified ject (3GPP) , especially for downlink CoMP, behind the gap between theoretical MU-MIMO IEEE Communications Magazine • February 2012 117 C qM IEEE M ommunications q qM Previous Page | Contents | Zoom in | Zoom out | Front Cover | Search Issue | Next Page MqM q Qmags THE WORLD’S NEWSSTAND® C qM IEEE M ommunications q qM Previous Page | Contents | Zoom in | Zoom out | Front Cover | Search Issue | Next Page MqM q Qmags THE WORLD’S NEWSSTAND® performance and that practically achievable so Some particular deployment scenarios may Energy efficiency is far in LTE deployments: emerge where the use of relays or repeaters may •Good spatial separation of users is most be beneficial (e.g., with very-low-power relays, an area of increasing easily achieved with correlated arrays of uni- outdoor-indoor coverage boosting, or on high- importance. All formly spaced antenna elements, whereas many speed trains), but in most general cases the loss future evolution of antenna configurations deployed in practice are of bandwidth on the backhaul link and the chal- cross-polarized, with low correlation between the lenge of interference control seem to make LTE-Advanced will polarizations. relays relatively unattractive for realizing useful therefore have to •CSI at the base station is far from perfect, capacity gains in LTE-Advanced systems in the due to the constraints of uplink feedback over- near future. give a high priority head and latency; channel reciprocity may be to such considera- exploited in time-division duplex (TDD) systems, OTHER DRIVERS tions. This will but the accuracy of the CSIT remains limited by factors such as non-reciprocal transmit-receive ENERGY EFFICIENCY involve minimizing antenna configurations, mobility of terminals, Although LTE has been designed from the the transmission and the accuracy of uplink channel estimation beginning to facilitate energy-efficient imple- achievable with reasonable uplink reference sig- mentations of both mobile terminals and base overheads and pro- nal overhead. stations, energy efficiency is an area of increas- cessing requirements •The availability of sufficient spatially separa- ing importance due to the rising cost of energy when traffic loading ble users with non-empty data queues is limited; and ecological concerns. All future evolution of nevertheless, as the LTE terminal population LTE-Advanced will therefore have to give high is light, and in increases exponentially, this will become less of a priority to such considerations. This will involve general enhancing difficulty, even when bursty traffic models are minimizing the transmission overheads and pro- assumed. It is shown in  that, for a given CSI cessing requirements when traffic loading is the ability of the feedback overhead, better MU-MIMO capacity light, and in general enhancing the ability of the network to adapt can be achieved with high-accuracy CSI feed- network to adapt dynamically to the actual load dynamically to the back from a small number of terminals (even levels. This may include more advanced tech- randomly selected) than with low-accuracy CSI niques for load balancing/transfer so that parts actual load levels. feedback from a large number of terminals. of the system which are not essential may be Future performance improvements with switched off whenever possible, while retaining downlink MU-MIMO will therefore depend on flexibility to bring them online rapidly to handle whether it is possible to significantly improve the high traffic demand when needed. accuracy of the CSIT, given practical channel estimation, realistic antenna configurations, and BACKWARD COMPATIBILITY typical terminal mobility, without incurring unre- One of the design principles of LTE has been alistic feedback overhead; some study of the backward compatibility between successive potential benefits of MU-MIMO enhancements releases of specifications. This implies that user relative to LTE-Advanced Release 10 has equipment of an earlier release can work in a already been carried out by 3GPP , but there network of a later release, and that user equip- remains room for further study in this area. ment of a later release can work in an older net- In the uplink, MU-MIMO can be implement- work. This is important since it ensures ed more easily because accurate CSIT is not continuity of service for existing subscribers and required. The complexity of the uplink MU- protects the investments of network operators. MIMO operation is managed at the receiver The requirement for backward compatibility rather than the transmitters, and sophisticated should not, however, prevent the introduction of non-standardized receivers, such as those based new features, since new functionality can be con- on MMSE-SIC, can theoretically attain the max- figured or enabled by the network specifically imum multiple access channel capacity . for the user equipment that supports it, without requiring all users to upgrade their equipment RELAYS immediately. Relay nodes provide an interesting example of In the future, for maximum efficiency, the small-cell-based heterogeneous networks, and requirement for backward compatibility will not much has been written about their potential. In be imposed on every LTE-Advanced carrier; general, the benefits of relays can be categorized provided there is always at least one carrier on into coverage extension (including filling cover- which legacy terminals can operate, new carriers age holes) and capacity boosting . The future may be aggregated for the sole use of newer ter- evolution of LTE-Advanced will be directed minals; sometimes referred to as extension carri- mainly at the latter, where the practical benefits ers, such non-backward-compatible carriers of relays appear to be still somewhat equivocal. could then exploit new features to the full, with- While significant capacity gains have been pre- out being constrained to carry all the signals dicted theoretically for systems using relaying (e.g., reference signals and control signaling) under some ideal assumptions, these gains are expected by legacy terminals; this can improve typically dependent on advanced “network cod- spectral efficiency by reducing overhead, as well ing” schemes that are complex and dependent as further facilitate energy-saving techniques on on high channel estimation accuracy. Particular the new carriers. attention also needs to be paid to the interfer- ence generated by relay nodes, which, without SELF-OPTIMIZATION careful control, can undo much of the gain that The deployment of a mobile communications results from the improved received signal power network is a complex task, especially its opti- delivered by the relay nodes. mization in different deployment scenarios and 118 IEEE Communications Magazine • February 2012 C qM IEEE M ommunications q qM Previous Page | Contents | Zoom in | Zoom out | Front Cover | Search Issue | Next Page MqM q Qmags THE WORLD’S NEWSSTAND® C qM IEEE M ommunications q qM Previous Page | Contents | Zoom in | Zoom out | Front Cover | Search Issue | Next Page MqM q Qmags THE WORLD’S NEWSSTAND® with variations in traffic patterns. The LTE sys- mechanisms to reduce cost is economy of scale; tem broke new ground with the way in which it therefore, excessive fragmentation of the LTE Mobile communica- was designed with certain self-optimizing net- chipset market may turn out to be counterpro- work (SON) functions built in from the start . ductive. tions traffic is no These functions include the facilitation of The delivery of multimedia broadcast/multi- longer dominated by deployment of new cells, with automatic detec- cast services (MBMS) over the mobile network traditional voice tion of neighbor cells, self-configuration of net- has attracted considerable interest in the last work interfaces, and automatic configuration of decade, and LTE is in a better position than any services, and it will cell identifiers. SON functions in LTE can also previous cellular system to offer such services in be important to optimize network operation by automatic load a spectrally efficient manner by exploiting the balancing, and by refinement of handover orthogonal frequency-division multiplexing continue to adapt parameters by automatic detection of handover (OFDM) signal properties for multipoint trans- LTE-Advanced to failures and identification of their causes (e.g., mission in a single-frequency network. The con- ensure efficient handovers that start too late or too early, or tar- sumer take-up of such services has thus far been get an inappropriate cell). very mixed, with large differences in popularity support for unique Further enhanced SON functionality, to aug- in different regions of the world. It remains to traffic characteristics ment the ability of an LTE-Advanced network to be seen whether new applications will emerge to reconfigure itself and adapt to changing scenar- inspire more widespread MBMS adoption in of applications that ios, will be an essential part of reducing the cost regions where it has not yet met with significant become relevant in per bit of operating the networks of the future. consumer confidence; if this were to occur, it the future. One tool likely to be increasingly developed in could provide motivation for further enhance- this area is automatic logging of measurements ment of LTE’s MBMS modes of delivery, for by user equipment in order to reduce the need example, to support higher data rates or more for manual “drive tests” to identify any network simultaneous service channels. coverage problems. Other areas on which it is worth keeping a keen eye are mobile gaming and active social SPECTRUM AVAILABILITY networking: the traffic characteristics of such The limited availability of suitable radio spec- applications are quite different from more con- trum will increasingly impact the future evolu- ventional voice or data services. Gaming applica- tion of LTE-Advanced. This is already evident in tions typically give rise to bursty traffic with a the carrier aggregation features provided by requirement for very-low-latency communica- LTE-Advanced, and it is inevitable that the tion; in some cases, accurate knowledge of the range of band combinations that have to be sup- location of mobile terminals may also be ported will continue to increase. Techniques to required. If consumer demand for such services enhance dynamic load management between takes off, it may be worthwhile to consider carriers according to traffic demand will also whether specific enhancements can support become an increasingly valuable tool for ensur- them in a more optimized way. ing full and efficient use of scarce spectrum The uplink-downlink traffic balance is another resources. factor that may impact the development of LTE- Such dynamic techniques are likely in due Advanced in the future. Historically, development course to evolve in the direction of cognitive of the uplink in most mobile communications sys- radio solutions, with increasing utilization of tems has lagged behind the downlink, in terms of spectrum sharing and white space detection as both market demand, and system specification spectrum becomes ever more crowded. and practical implementation. Mobile networks have typically tended to experience downlink- heavy traffic loads; nevertheless, applications such NEW APPLICATIONS AND as file sharing in the fixed Internet have resulted in a more even balance, and this may perhaps be TRAFFIC PATTERNS seen as an indicator of future trends in mobile Mobile communications traffic is no longer dom- networks. It is likely, therefore, that in the future inated by traditional voice services, and it will be further attention will need to be paid to enhanc- important to continue to adapt LTE-Advanced ing the uplink performance to more closely match to ensure efficient support for unique traffic that delivered by the downlink. Constraints to be characteristics of applications that become rele- overcome include the lower transmit power of vant in the future. mobile terminals, which results in uplink coverage Traffic in mobile communications networks is limitations, and the cost of mobile terminals. no longer solely generated by traditional mobile More advanced multiple-antenna schemes than terminals. Increasingly, mobile networks are the four-transmit-antenna single-user MIMO being used to gather data from or transfer data techniques standardized in LTE-Advanced between machine-type devices. Such communica- Release 10 may provide one route to address tions have certain particular characteristics, typi- these challenges. cally including small packet size, tolerance of transmission delay, low mobility, low frequency of use, a need for long battery life, and poten- IMPLEMENTATION ADVANCES tially a very large number of terminals; the ter- Not all enhancements to LTE-Advanced will minals themselves need to be very low in cost. require changes to the system specifications. In Specific features to optimize the support provid- parallel with the introduction of new features in ed for these types of communications may be the specifications, implementation advances are introduced in the future, although it should be also to be expected, as processing capability borne in mind that one of the most effective increases and new algorithms emerge. For exam- IEEE Communications Magazine • February 2012 119 C qM IEEE M ommunications q qM Previous Page | Contents | Zoom in | Zoom out | Front Cover | Search Issue | Next Page MqM q Qmags THE WORLD’S NEWSSTAND® C qM IEEE M ommunications q qM Previous Page | Contents | Zoom in | Zoom out | Front Cover | Search Issue | Next Page MqM q Qmags THE WORLD’S NEWSSTAND® ple, algorithms for channel estimation or inter- of the LTE and LTE-Advanced standards, as It is clear that ference cancellation that once seemed infeasible well as his colleagues Federico Boccardi, Fang- at reasonable cost will tomorrow be entirely pos- Chen Cheng, Howard Huang, Sudeep Palat, LTE-Advanced will sible. Philippe Sehier, Antonia Tulino, and Sivarama need to continue to On the UE side, such advances are likely Venkatesan, and the anonymous reviewers, for pursue bold from time to time to result in new, tighter per- their helpful comments in the development of formance requirements being standardized to this article. evolutions and identify superior terminals. embrace new On the network side, some features have REFERENCES been highlighted above that can benefit from  S. Sesia, I. Toufik, and M. Baker, LTE — The UMTS Long technologies to keep new network architectures with centralized pro- Term Evolution: From Theory to Practice, 2nd ed., pace with market cessing; significant effort will also be directed Wiley, 2011.  S. Venkatesan, A. Lozano, and R. Valenzuela, “Network demand and remain toward enhanced algorithms for scheduling and MIMO: Overcoming Intercell Interference in Indoor coordination between cells and between trans- Wireless Systems,” Proc. 41st Asilomar Conf. Signals, the leading global mission/reception points. Such algorithms will Systems and Computers, Pacific Grove, CA, Nov. 2007, not be standardized, but will have a major impact pp. 83–87. mobile  S. A. Ramprashad and G. Caire, “Cellular vs. Network on the performance and reliability offered by the MIMO: A Comparison Including the Channel State communications LTE-Advanced networks of the future. Information Overhead,” Proc. IEEE PIMRC ’09, Tokyo, standard. Japan, Sept. 2009, pp. 878–84.  3GPP, “Technical Report 36.819: Coordinated Multi- CONCLUSIONS Point Operation for LTE, Physical Layer Aspects,” www.3gpp.org. It is clear that LTE-Advanced will need to con-  K. Balachandran et al., “NICE: A Network Interference tinue to pursue bold evolutions and embrace Cancellation Engine for Opportunistic Uplink Coopera- new technologies to keep pace with market tion in Wireless Networks,” IEEE Trans. Wireless Com- mun., vol. 10, Feb. 2011, pp. 540–49. demand and remain the leading global mobile  S. Venkatesan, “Coordinating Base Stations for Greater communications standard. Uplink Spectral Efficiency in a Cellular Network,” Proc. Meaningful enhancements are likely to focus 18th IEEE PIMRC, Athens, Greece, Sept. 2007. on techniques that improve the uniformity of  M. Costa, “Writing on Dirty Paper,” IEEE Trans. Info. Theory, vol. 29, May 1983, pp. 439–41. service provision by boosting cell edge data rates  N. Jindal and S. Ramprashad, “Optimizing CSI Feedback as well as overall system capacity. for MU-MIMO: Tradeoffs in Channel Correlation, User Processing in the network will increasingly Diversity and MU-MIMO Efficiency,” Proc. 73rd VTC, become centralized and ultimately cloud-like, Yokohama, Japan, May 2011, pp. 1–5.  3GPP, “Technical Report 36.871: Downlink MIMO taking a multicell or system-wide view of net- Enhancement for LTE-Advanced,” www.3gpp.org. work optimization in order to manage interfer-  H. Huang, C. B. Papadias, and S. Venkatesan, MIMO ence and increase overall capacity and flexibility. Communication for Cellular Networks, Springer, 2012 Network adaptability will be a key focus, with  F. Boccardi et al., “An Industrial Perspective of Relay- ing for Cellular Systems,” Proc. 44th Asilomar Conf. the aim of responding dynamically to changes in Signals, Systems and Computers, Pacific Grove, CA, traffic patterns and reducing the operating costs Nov. 2010, pp. 1311–15. of the network; enhanced SON functions will play a key role in such dynamic optimization. BIOGRAPHY Adaptability will also extend to flexible exploita- MATTHEW BAKER (firstname.lastname@example.org) is the ___________________ tion of available spectrum as radio frequencies Chairman of 3GPP TSG Working Group 1. He holds degrees become increasingly crowded. in engineering and electrical and information sciences from Finally, the characteristics and requirements the University of Cambridge, United Kingdom. From 1996 of emerging applications and device types must to 2009 he worked at Philips Research where he conducted research into a variety of wireless communication systems be carefully monitored and assessed to ensure and techniques, including propagation modeling, DECT, that LTE-Advanced has the necessary capabili- HiperLAN, and UMTS, as well as leading the Philips 3GPP ties to meet their challenges. RAN standardization team. He has been actively participat- ing in the standardization of both UMTS WCDMA and LTE ACKNOWLEDGMENTS in 3GPP since 1999. He has been with Alcatel-Lucent since 2009. He is a Chartered Engineer, a member of the Institu- The author gratefully acknowledges the work of tion of Engineering and Technology, and a visiting lecturer all participants in 3GPP toward the development at the University of Reading, United Kingdom. 120 IEEE Communications Magazine • February 2012 C qM IEEE M ommunications q qM Previous Page | Contents | Zoom in | Zoom out | Front Cover | Search Issue | Next Page MqM q Qmags THE WORLD’S NEWSSTAND®