IEEE Wireless Communications | My Assignment Tutor

40 1536-1284/18/$25.00 © 2018 IEEE IEEE Wireless Communications • August 2018AbstrActMmWave communications applied to smallcells has been recognized recently as an important means to break the spectrum gridlock andto dramatically scale up the system capacity forthe evolving 5G mobile networks. However, significant technical challenges must be resolved tofully exploit the potential of mmWave small cells.Among them, base station discovery is particularlychallenging due to the severe propagation loss atmmWave bands. Beam alignment arises as anothercritical problem that must be solved adaptively for5G mobile access. This article discusses protocolsand techniques for these two aspects, and presents important system design insights. Some openresearch problems are also highlighted in base station discovery, multi-user beam alignment, and network management.IntroductIonThe evolving fifth generation (5G) mobile networkis facing many-fold fundamental requirements, asdriven by an increasingly diverse range of mobileservices, such as multimedia, virtual reality, theInternet of Things, and wireless automotive applications. The 5G mobile network is anticipatedto support over 1 million connected devices persquare kilometer, a 1000-fold increase in systemcapacity (bits per second per unit area) comparedto 4G, up to 10 Gb/s data rate links, and extremely low latency (1 ms), all with very low infrastructure deployment cost [1].Millimeter-wave (mmWave) communication isan important means to dramatically scale up thesystem capacity for 5G mobile networks due toits abundant frequency in the range of 30–300GHz. The feasibility of mmWave cellular accesswas recognized by industry as early as 2011 andhas been validated by various channel measurement campaigns in outdoor environments [2, Sec.III]. Although there are still worldwide debates onwhich band 5G should be using, many countriesare now racing to explore mmWave spectrum. TheThird Generation Partnership Project (3GPP) is alsoworking toward standardizing mmWave for the 5GNew Radio (NR) interface [2–4]. Recent progressincludes channel modeling for radio above 6 GHzand some agreements on the high-level design ofthe physical layer. However, detailed solutions andspecifications remain open.Due to severe path loss and sensitivity to blockage, mmWave communication for long-range cellular access is problematic, but it provides greatopportunities for small cells, which target supporting short-range communications in densely populated hotspots. The fusion of mmWave and smallcells provides both large bandwidth and high spatial reuse, and is expected to be one of the keyenabling technologies to support high-data-ratetransmission in 5G.Figure 1 depicts a typical deployment ofmmWave small cells overlaid with a conventionalwide-coverage LTE/5G macrocell. As illustrated,mmWave small cells are anticipated to accommodate high-volume data traffic demand from indoor/outdoor hotspots, to provide high-speed cellularaccess/backhaul services, and also to supportemerging automated driving applications, wherehigh-definition sensing data (e.g., maps) is transmitted among vehicles, or between vehicles androadside units [5].Compared to conventional LTE/5G (sub-6GHz) small cells, mmWave small cells are moreeffective in scaling up the system capacity and supporting multi-gigabit-per-second data transmission.Simulations [6] have shown that it is possible toachieve 59 Gb/s/cell throughput (about 1 Gb/s/user equipment [UE] for 60 UE/cell) for 50 msmall cell size, which leads to 1000 times per-cellthroughput increase over a typical LTE/5G smallcell.MmWave small cells require much simplifiedinter-cell interference management. They operateat a frequency band different from macrocells, thuseliminating interference management betweensmall cells and macrocells. Inter-cell interferencebetween mmWave small cells is less significant,due to the adoption of directional transmission andsevere path loss at mmWave. In addition, indoorand outdoor mmWave small cells would hardlyinterfere because of the high penetration loss.MmWave small cells can naturally integrateaccess and backhaul using the same frequencyband. This eliminates the need for wired/fiberbackhaul as in the conventional small cells, whichis usually costly and location-constrained, and alsotakes time to install. MmWave wireless backhaulalso allows more flexible resource management,which can be adapted to traffic demand.In spite of these promises, mmWave applicationto mobile networks is still in its infancy and facesChunshan Liu, Min Li, Stephen V. Hanly, Philip Whiting, and Iain B. CollingsMillimeter-Wave Small Cells:Base Station Discovery, Beam Alignment, andSystem Design Challenges5G MMWAVE SMALL CELL NETWORKS: ARCHITECTURE, SELF ORGANIZATION, AND MANAGEMENTThe authors are with Macquarie University. The corresponding author is Min Li.Digital Object Identifier:10.1109/MWC.2018.1700392Authorized licensed use limited to: Macquarie University. Downloaded on March 31,2020 at 04:55:01 UTC from IEEE Xplore. Restrictions apply.IEEE Wireless Communications • August 2018 41significant technical challenges. In particular, therequirement of directional transmission is a keydriver for innovative system designs for mmWavesmall cells. At the physical layer, research effortsare needed to better understand how to discover,establish, and maintain good-quality links for directional transmission in mmWave small cells.Base station (BS) discovery (a.k.a. cell search)is a crucial step in establishing a communicationlink in any cellular system. It allows a UE to detectthe presence of a BS and extract relevant timinginformation. In conventional LTE systems and theproposed 5G at sub-6 GHz, this involves the BSsperiodically broadcasting synchronization signals(also referred to as reference signals [RSs]) in anomnidirectional manner and UEs scanning for thepresence of the signals to discover and synchronizeto a BS. But due to the unfavorable propagationat mmWave, such omnidirectional transmissionmay not suffice to deliver enough power to ensurequick discovery of BSs at UEs. This makes beamforming transmission of RSs important for efficientBS discovery.Design of new beamforming-based mechanismsfor mmWave BS discovery has drawn significantattention from both academia [7–9] and the 3GPP[3]. Barati et al. [7] studied the performance of BSdiscovery with directional transmission of RSs viarandom beamformers. We subsequently proposeda sequential beamforming strategy and developeda few fundamental limits on the performance ofBS discovery [8]. As an extension to [7], Barati etal. [9] looked at the whole initial access procedure(including BS discovery) and evaluated the overallaccess delay under different BS/UE beamformingconfigurations.Successful BS discovery and synchronizationare prerequisites for the subsequent data transmission phase, in which beamforming also plays a vitalrole. To achieve large beamforming gain, transmit/receive beams at the BS and UE must be adaptively steered and aligned. One viable approachfor beam alignment at mmWave is beam trainingthrough spatial scanning, in which the BS and UEjointly examine BS/UE beamforming pairs frompre-designed codebooks that represent the beamsearch space to find strong multi-path components.This approach does not require explicit estimationof the channel, and is particularly useful to identifythe dominant path in the sparse mmWave channeland align the transmission and reception along thispath.A conventional strategy for beam training isto perform an exhaustive search by examining allbeam pairs in the codebook and determining thebest pair that maximizes a given performance metric (e.g., beamforming gain). The training overheadof this strategy is proportional to the size of thebeam search space and thus can be prohibitivewhen narrow beams are employed. To reduce thetraining overhead, [10, 11] considered a hierarchical search based on multi-level codebook designs.Our recent study [12] established fundamentallimits in beam alignment performance under bothexhaustive search and hierarchical search subjectto the same training overhead. The 3GPP proposal[4] discussed several beam sweeping mechanismsthat take into account different system perspectives including load balancing among beams, beammanagement complexity, and feedback signalingoverhead reduction. A Markovian interactive beamsearch was proposed in [13] after reasonable simplification of mmWave channel models. While [4,10–12] investigated beam training at the link level,[14] looked at the problem in a multi-cell network.In this article, we review our proposed protocol for mmWave discovery and discuss some fundamental design insights in the following section.We then present two classical beam alignmentapproaches and discuss their fundamental performance limits. Finally, we highlight several designaspects toward multi-cell multi-user mmWave smallcell systems, inspired by the results and insights inmmWave BS discovery and beam alignment.mmWAve bs dIscovery WIth beAmformIngIt has been well recognized that beamformingtransmission of RSs is important for efficient BSdiscovery. When beamforming is adopted, however, spatial scanning is inevitable to ensurereasonable discoverable range to UEs in all directions. Suppose multiple beams are multiplexedin time to provide universal coverage in all directions, and that the total overhead (the percentage of time used for RS transmission) is fixed. Ifnarrower beams are deployed, higher beamforming gain can be achieved in each beamformeddirection; on the other hand, since more beamsare required to cover all the directions of interest, a smaller fraction of time is made available toeach beam. Thus, it is unclear what the impact ofdifferent beamforming schemes is on the overallperformance of BS discovery.In this section, we first review our proposedprotocol for mmWave BS discovery, and then discuss key insights on the roles of beamforming andother system parameters on the performance of BSdiscovery [8]. We use both theoretical and simulated results to unveil the value of topology information of mmWave BS coverage.Protocol for mmWAve bs dIscoveryThe protocol is based on a generalized LTE-typeframe structure [7, 8], as depicted in Fig. 2. EachBS broadcasts its RS in the 0th slot of every frame.The RS is of duration Trs < Tslot


Leave a Reply

Your email address will not be published. Required fields are marked *