Journal of Communications and Information Networks Vol.1, No.2, Aug. 2016 @>‚›*ƒ**¤¤Œ%ƒ¤+*ƒ‡*%ƒ*+ƒ††Technical challenges for high-frequency communication Review paper

Technical challenges for high-frequency wireless communication

ZHAO Jianping, LI Chenlei, LI Xianghua, SHEN Long

Huawei Technologies Co., Ltd., Shanghai 201206, China

Abstract: Recent rapid developments in wireless communication have been motivated by breakthroughs in {`@"=`@Q"=#>?@"> ?_Q"{#=|?|?_#N Q _NQ Q_ }}Q# = Q‚\Q|? summarize technical challenges ranging from propagation attenuation and the implementation of circuit devices, to signal ¡\Q

Key words:_}¡

Citation›‰|=>£„}‚}}‚|_˜£™£ %ƒ*+*%#›*¤%‡

1 Introduction {*ƒƒƒ capacity, facilitate connections for at least 100 billion `@"= Q*ƒŒ{ >?@"QŽ capable of extremely low latency and response 2009, the 4G wireless communication industry has times˜†™‚Q_ been on the verge of remarkable developments. ‚‚# `†„„ !\!\# Release 12 features, which represent 4G+ systems[1], ›ŒŒŒ%ƒ*+Œƒ%Œ%+Œ%ƒ*+ seems premature at present, a revolution in wireless Q# network technology is on the horizon[2], due to ?|?+| enhancements in future releases as well as the wide +_+|# due to wide signal bandwidths of up to several GHz, {%ƒ%ƒ the focus on beam, and excellent direction %ƒ†ƒ ‚|?

"Q=&%ƒ*+¥=*&%ƒ*+ 20 Journal of Communications and Information Networks

provides high data rate given rapid mobility. has that is scores of times wider than 4G œ & as well as a higher scale of antenna array, data rate is interface technology. There are a number of candidates higher and the computational burden due to digital Q?>?@"?>?@"# Q? "="=#„[4], __ ?'"?'"#?@" combined, the system’s algorithms need to exploit the ?_Q"{#œ?"œQ advantages of both. All the above problems need to "#˜™Q>?@" Q|?Q "Q ‚\Q|? are unclear in this context, no agreement has been technology, summarize the key technical challenges, _ ¡ i.e., macrocell, microcell, picocell, and at example. _%‡|†‡|+ƒ|”ƒ¦ GHz are all still options. As a result, the protocol for 2 Propagation attenuation ŽQ view holds that there is a large gap in current LTE 2.1 Comparison of wireless channel between _ HF and sub- To bridge this gap, a combination of a wider @Q{ spectrum, higher spectral efficiency, and decuples "`|? of higher network density is necessary. However, + _ ? Q›*# \ „›? different weights. The measured data has shown that |?"`Q# |? is a higher probability of collision with particles in dominance. Some aspects of a comparison of wireless the air than bypassing them, attenuation is higher. |?+› ? |? *#} }}# ‚+} Ž}Ž# approximate to that in free space. Along the NLoS should be reconstructed using a large amount of „}„}{#‚ %#‚ |?} ‚?Œ\?‚Œ\?_#› Ž}„}? ?{|?„=„=# {Q|? has limited output power and linearity, whereas the _?* |?Q Q+ "Q|? `+ +†# standard deviation is relatively small and the shadow {› area is large, such that the classification of the Technical challenges for high-frequency wireless communication 21

?|?Q relatively large and the shadow area small, such that =‚\„§&ƒ' spectral efficiency is 4.6 bit·s1·Hz1, the coverage }Ž}Q‡ *ƒ¥ƒ+^1·Hz1, the coverage distances of LoS and NLoS are 406 m &”Ž} }? Q‚\„ coverage distance, which is demonstrated by the Q?%?{ ƒ+^1·Hz1, the coverage distances of }Ž}+†‡ ”†Q %# ?*"Ž} ‚+ outdoor microcell scenario. |?Q "Q„\„ the LoS, plays a much more important role in \#|?Q Q\ „\„\# increase, and the scattering feature is discrete } ›|?"` `„@„ +Q „@„#|? œ¡  as an example, we show the link budget curves in arising from the power against the delay becomes ?%› QQ Q?% ? %‡| the delay spread decreases while angular spread red and blue curves, respectively, correspond to the increases for the sparse property.

?%} ¡›#}¥#Ž} 22 Journal of Communications and Information Networks

†#? devices +|? # › channel clusters and a smaller spread of a single |?{ ? researched, since the relevance of devices and the channel clusters, and are closely related to the channel features is strengthened due to the circumstance and distribution of scatterers. The Q{ channel clusters show a stronger mapping to the # |› + œœ_# 2.2 Requirements of channel measurement # }Q and modeling Q›@%@ "%" *# this channel under different scenarios.  {& ‚†& _|?› ?{ patterns with directional antennas, and considering Q_ the diversity of the antenna configuration, the features, it has broken the conventional frameworks. performance of the transmission technology system, Therefore, it is necessary to design measurement {"Q{ Q_ Q Q{ be tested by implementing the antenna model with feasible modeling schemes under the conditions of %#\_ |?_ |? \ª@Q# › algorithm design. # ›Q the channel due to the mobility of users 3 Implementation of circuit devices # "Q›Q *@%@Q†@  antenna arrays _‚\„Q‚ # Q› \„#Q scatterer, and an extended statistical model of |?|Q the scatterer’s distribution QQ‚\„ # †@›†@ Q necessary, but leads to problems in the design and channel with smaller cell coverage \?‚\?_ # " › ‚#? multiple coexisting channels due to the rapid |? increase in the number of mobile users and and the circuit. Technical challenges for high-frequency wireless communication 23

3.1 High-frequency antennas \?‚‚„{ but offers only a slight advantage in terms of output #  power over SiGe. Heterogeneous integration has _*Œ*ƒQ*Œ*ƒƒ {_\?‚? + {‚‚‚! |? ">„= „'| with LNA. However, the process of the development \? of heterogeneous integration is still not mature. The greatly reduced, so that cost can be reduced. = # |? _\?‚{Q problem because the „=„= Q_‚ of the LNA, and the insertion loss of the switch, due  to its high electron mobility, high breakdown voltage, ? Q_Q[6]. the problem of electromagnetic compatibility. However, the cost of this process is much higher As a result, the homogeneity of the antenna "> pattern deteriorates. Q # Q ">Q \? which results in insertion loss. Therefore, it is _\?‚ a key technology to design an ?>‚ approach with low insertion loss. {‚ # ‚ and exhibits a considerably smaller insertion loss appropriate material for an antenna’s radome QQ"> and an appropriate manner of installation, as _

|? The ft and fmax of the transistor indicate the are unknown. {_ Q_ 3.2 The fabrication of high-frequency RFIC Q?†

illustrates the ft and fmax">>‚

`_\?‚ processes. As can be seen, an ft*†ƒ _Q &>‚&ƒ">%ƒ| Q and these three fabrication processes are sufficient Q\?‚Ž‚„= ¡!

">>‚ \?‚?ft of 90 nm SiGe =Ž„={{ %‡">Œ>‚†ƒƒ| „=„ `\?‚ =#_ _\?‚\? %ƒ|_ lower power consumption. 24 Journal of Communications and Information Networks

the future.

3.3 Integration of antenna array and RF module

@{ _ in current cellular , and limits the distance of Q_|Q \?QQ› *#„=

?†ft ŒfmaxN „=%# from globalfoundries for example, enhancing the scale of the antenna array ' _\?‚ _Q smaller, and the size of the antenna array becomes The cost of the SiGe process is lower than that of the GaAs, but is still higher than that of the SOI and the elements is set to a half the wavelength to reduce the ">"> # SiGe exhibits satisfactory phase noise, output power, \? „=!>„=Q \? concentrated research efforts have drastically reduced module can minimize the size of the wireless system "> \?‚ "> antenna elements. There are three ways to implement _\?‚ \? „=„= ?&˜”™›*#=„= }Ž=Q „'#›\?‚„' ">>‚ ¥%#=„= has the comparable performance on active devices. „ #›\?‚\ However, the insertion loss in passive components which tiles with an embedded antenna. The tile >‚"> „' and SiGe. Hence, the SOI process is a good choice '='=#¥†#==#› Q the antenna is implemented directly on a glass wafer Q„= \?‚ LNA, switch, etc. However, there have been a few =„ reports concerning the modulator, the demodulator, !>!>#„}} insertion loss degrades as the number of layers „} }# „'Q >‚\?QR QQ„' SOI needs to be further researched and explored in "Q Technical challenges for high-frequency wireless communication 25

compatible with the mainstream manufacturing and %# {\? †#Q _\?‚ degrade antenna performance˜‡™. The AoC integrates \? same chip in silicon technologies. This is usually |“*ƒƒ|#Q wireless communication, and array applications. elements of AoC is much lower due to large Ohmic losses and surface waves[9]?"‚ ‚# =„= and the circuits.

3.4 Challenges arising from large bandwidth

The Shannon–Hartley theorem states that channel capacity is proportional to the bandwidth of a communication system. In other words, the data rate can be increased by widening the bandwidth of the |Q+| are precious and the bandwidth of current cell %ƒ"|? _Q Q attain several hundred megahertz or gigahertz of *ƒ“*ƒƒ }_ *ƒ“*ƒƒ*ƒŒ ?& |QQ_ \?›#=„¥#=„; #= wide bandwidth, there are several technical challenges in the design of circuit components high and the maturity is unsatisfactory. Although Q Q=„ wave communication[10]?{ QQ„' ‡ƒ%**%*+| the complexity of system assembly increases. circuit components and antenna should cover the ?*# \?”“+&|# 26 Journal of Communications and Information Networks

\? a constant modulus. "Q Analog has been proposed by _\Q Q be considered[11]. consisting of beam pattern generation using a ? interference is not dominant, and the thermal noise at the receiver, and feedback from the selected of the system is comparable or even larger than that beam indices˜*†*&™ +[12]Q|?Q advantage of without the prior channel knowledge at degrades, the output power density per of the the . However, it consumes large overhead system decreases as bandwidth widens, and the range for the feedback scheme, which results in outage. of coverage shrinks. "Q into account the digital baseband. 4 Since the analog process can significantly reduce the complexity of signal processing, it is a better =Q method to combine digital and analog beamformers. antenna arrays and wide bandwidth are key features of "Q |? obtained by the receiver sounding reference signal, the amount of storage and the burden due to digital according to channel reciprocity between opposite computation. Therefore, more advanced architectures @@ _ =%*|? _Q the channel has the sparse property of multipath Q› scattering, such that channel estimation can be sparse Q"‚">{ 4.1 Low-complexity beamforming with sparse elements of the main channel scattering = }Q |'? Q propagation paths being selected at the beamspace, Ž\Ž\#? which are further digitally combined at the baseband large arrays may enable the precoding of multiple and can be solved through compressed sensing˜**+™. data streams, which can help improve spectral |? channel should be measured and demonstrated, and ?+"‚">"‚ the computational burden of compressed sensing "># Q‚|'? digitally at the baseband, where the phase and long way to its application. amplitude of the signal are controlled. However, _ 4.2 RF channel calibration \? which is costly in terms of power consumption for =Q |? a significant means of increasing the capacity of Technical challenges for high-frequency wireless communication 27

?Q"‚"> ?_ |'? channel. \?œ \? manufactured perfectly, and the difference in the Q\? Q amplitude error, and the phase error commonly exist. ‚|?Q common proposal has been proffered for air interface design, a distinct trend is that each resource element ??_ = _\? |?+ + bands reduces overhead, and even captures the of phase error calibration can be lower, whereas the _ Qœ on the extent to which overhead is acceptable and the higher. However, a dedicated calibration algorithm extent of overlapping of coverage of both subsystems. |? 5 Conclusion and future research 4.3 Coordination of low-frequency and high- frequency systems |?"` +|? Q|? QQ large bandwidth. However, the overhead involved in the beamforming process is large. An alternative features in wireless channels through measuring and _ ? _ |?Q QQ |?\?‚¥ _ Q|? œ|? for both improving the system capacity and degrading Q _|? _œ devices. Therefore, the implementation of the forthcoming Q|? |? Q˜*”™. The third Q"‚">{ QQQ_ loss, sparse recovery, and channel prediction. subsystem of the initial value of the channel at a Q|? References given detected values of both subsystems to give a predicted value of the next snapshot. This process, ˜*™\>}\=|}‚Ž‰£"\¡}}Q 28 Journal of Communications and Information Networks

†„„\*%#˜\™"› ˜£™„‚%ƒ**¤¤‡#›*†¤ƒM =Ž*"=%%\ª‚ *&†+ ˜%™"=}}‚Ž>Ž¡›QQ ˜**™ „>\‚Ž>@|‚\`œ› ˜'Œ>}™›ŒŒ†ŒQŒ†Œ*””&® ˜£™‚%ƒƒ† wiseharbour. &*”#›++”& ˜†™ ›=!˜'Œ>}™|%ƒ*† ˜*%™\=Ž=Ž\=„„=„>\\¡‚„"Q ˜&™›Ž=‚\=!˜'Œ>}™ ›˜£™„ |%ƒ* ‚%ƒ*&*ƒ%†#›†++†‡ ˜™Q_˜\™@› ˜*†™‚„‡ƒ%**„**›`}=Ž"= <%ƒ* "=#„}„|š#=†› ˜+™@>=Ž|"="‚Ž‚¡Ž£=@=""Q Enhancements for Very High Throughput in the 60GHz band[S], ">˜£™‚%ƒƒ&ƒ*#› International Standard, 2012. *&&* ˜*&™|œ\¡‚"}>!@£"Q ˜”™œ!=}@=\‚==Ž==\=£=Ž=` for wireless backhaul and access in networks[J]. IEEE scalable phased arrays for imaging and communications[J]. IEEE %ƒ*†+**ƒ#›&†¤*&&ƒ† %ƒ*†&#›*¤+%ƒ& ˜*™}£‚}}š{ ˜‡™¡="@}‚œ@Ž==\=£=Ž= "‚">Q +ƒ|˜£™ ˜™ŒŒ„=œ=%ƒ*&› IEEE transactions on components, packaging and manufacturing ††%+†††* %ƒ*****#›*‡ƒ+*‡*& ˜*+™=š=|>\=£=>„=}='œœ\= ˜¤™‰|=Žš„}‚œ@= Q"‚">˜£™‚ QQ %ƒ*&*††#›*&¤¤**† communications[J]. IEEE transactions on antennas and propagation, ˜*”™Ž‚|?}>\=¡Ž‚|}š %ƒƒ¤”*ƒ#›%‡†ƒ%‡&* ›`Q'‚'"˜™ŒŒ ˜*ƒ™\=„„=„>\"œ\@>¡£Žœ‚\\‰? „‚ +ƒ| ‚Ž?>>"#|¡%ƒ*›%&*+%&%&

About the authors

ZHAO Jianping *¤”ƒ|Q LI Xianghua *¤‡”|Q' „@‚„ „@ 'œQ„ in physical electronics from Southeast University, *¤¤¤||} Ž%ƒƒ¤%ƒ*Q| %ƒƒ‡{= has been a research engineer at Technologies \?| }"%ƒ* \? _ ›± _ # Q›{*±#

LI Chenlei *¤‡†|Q„@ SHEN Long*¤‡‡|Q in Information and Communication Engineering " |‚"%ƒ* œQ"%ƒ*&| |‚„ \ª@|} and Geoscience & Remote Sensing Society, and a "%ƒ*& reviewer of IET and Elsevier publication. He has _ been a research engineer at Huawei Technologies ›± }"%ƒ* # Q"‚">_ ›± #