AN OPTIMIZED HYBRID BEAMFORMING FOR MILLIMETER WAVE MU- MASSIVE MIMO SYSTEM

Transcription

1 An optiized hybrid beaforing for illieter wave u-assive MIMO syste AN OPTIMIZED YBRID BEAMORMING OR MILLIMETER WAVE MU- MASSIVE MIMO SYSTEM Saeed Ali Saeed Alquhaif University of Lahore, Lahore, (Pakistan) E-ail: Ishtiaq Ahad University of Lahore, Lahore, (Pakistan) E-ail: Mahnoor Rasheed University of Lahore, Lahore, (Pakistan) E-ail: Ali Raza University of Lahore, Lahore, (Pakistan) E-ail: 93

2 An optiized hybrid beaforing for illieter wave u-assive MIMO syste ABSTRACT Millieter wave (Wave) and assive ultiple-input ultiple-output counications (assive MIMO) have been adopted as e ost iportant keys for e next generation (5G) wireless networks due to high spectru availability and assive use of space diversity for high roughput counications. Massive MIMO ust be ipleented parallel wi Wave to overcoe high pa-loss issue in Wave band by iproving e antennas gain. Digital precoding in assive MIMO shows high perforance at e cost of hardware coplexity due to e requireent of huge nuber of radio frequency (R) chains and analog-to-digital converters (ADCs). On e oer hand, analog precoding has less coplexity wi liited perforance as it supports only one data strea. In coproise between ese two, a hybrid precoding for ultiuser assive MIMO (MU-assive MIMO) systes wi low-coplexity and reasonable perforance becoes necessary for next generation networks. We propose an optiized hybrid precoding and cobining schee for fully connected structure. The design of cobined analog and digital precoding for e Wave MU-assive MIMO syste becoes a non-convex proble. Therefore, we proposed a suboptial design, where we split e transitter and receiver hybrid precoding and decoding and adopt orogonal atching pursuit algori for eir design. The siulation results show at e perforance of proposed hybrid precoding approaches e digital precoding in e ter of total su data rate as we increase nuber of R chains and SNR. KEYWORDS ybrid precoding, beaforing, MU-MIMO, Massive MIMO. 1. INTRODUCTION Scarcity of spectru has always been a proble in counication networks. Researchers always look adequate band for good propagation properties. Unfortunately, e current spectru does not fulfill e need of bandwid greedy broadband counications. or future wireless syste such as 5G, data rates for a user expected to increase explosively. Therefore, e researcher is working to ove eir attention to illieter wave (Wave) counications fro icrowave counications due to e availability of a huge aount of unused spectru at Wave. Counication at Wave is targeting a speed in Gbits/s (data rates about 10 Gbits/s, which is ore an current 4G 100 ties) according to e IEEE 80.11and WiGig. owever, operating at Wave coes wi e high signal attenuation and scattering liitations. To overcoe e issue of signal attenuation any researchers have proposed a new sall cell syste like icro-cell, feto-cell and pico-cell. By e sall cells, we can cobine Wave and MIMO syste to perfor 3D beaforing easily. Directional counication which requires transit and receive antenna beaforing can iprove signal-to-noise ratio (SNR) to e sufficient level in e Wave bands. Ipleentation of e assive MIMO at Wave for beaforing enables us to install hundreds or ousands of antennas at e base station and receiver, which increases SNR at e receiver, consequently, a significant increase in syste capacity can be observed. Wi digital beaforing, ultiple streas can be transitted in parallel, however it requires large nubers of radio frequency (R) chains and analog to digital converters (ADCs) which akes e digital beaforing ore coplex and costly in e ter of power consuptions. While analog precoding has less coplexity and can be ipleented wi e phase shifter at controls e phases of transit signal, but it supports only one data strea wi high SNR, resultantly e capacity cannot be increased significantly. owever, cobining of bo e schees 94

3 An optiized hybrid beaforing for illieter wave u-assive MIMO syste analog and digital precoding, enables us to get e advantages of e bo. Therefore, hybrid beaforing becoes a hot research issue [1]. While designing precoding schees for Wave assive MIMO syste, ostly researchers consider fully connected structure []. Many algoris are used for different structures for hybrid precoding. Assuing perfect channel knowledge at e transitter, a low-coplexity iterative algori based on orogonal atching persuit (OMP) is proposed in [3]. The hybrid technique provides a trade-off between e analog precoding at offers low-coplexity wi liited-perforance and digital precoding having high coplexity wi good-perforance [4][5]. One of e ost advantages of hybrid schee is at it can deal wi e less nuber of R chains and ore nuber of antennas [6]. In [7], e spares nature of e Wave channel is studied. ollowing [6], e auors in [] designed hybrid precoding algori based on alternative iniization and atrix decoposition. In [8] syste odel investigates bo fully connected structure (each antenna is connected to all R chains) and partially connected structure. Based on spatially sparse channel, a hybrid precoding is proposed in [9] at solves e proble by spares approxiation approach for achievable rate as e nearoptial solution. A codebook based algori is proposed in [10] at involve an iterative search in predefined codebook to get e atrix of optial hybrid precoding. In e existing research, e ost works on hybrid beaforing are focusing on single user MIMO (SU-MIMO) systes and trying to reduce e R chains wiout affecting e ulti-data streas. It is found at, e nuber of R chains are still ore an data streas, which akes e hybrid beaforing hard to be ipleented and costly in e ter of e power consuption. In our work, we proposed a novel hybrid precoding where a group of antenna serves a group of users siultaneously as a result e required nuber of R chain are reduced. We design hybrid precoding for ultiuser assive MIMO (MU-assive MIMO) in Wave syste as follows: (i) irst we design analog and digital decoder for each user. (ii) Use effective channel to calculate analog precoder, which is sae for all e users. (iii) Adopt block diagonalization (BD) for itigate e interference and intrainterference between e users. (iv) By soe available algoris, we design analog R precoder syste. (v) Digital precoding designed by BD axiizes e total rate for ultiuser assive MIMO syste in downlink channel. (vi) ybrid beaforing has been design by weighted su ean Square error (WSMSE) for which is e inverse gain of its digital precoding. (vii) inally, we study e perforance of bo digital and proposed hybrid beaforing wi different paraeters like SNR, data streas and R chains. The rest of e paper is organized as follows. Massive MIMO and Wave counications are studied in detail in Section. Syste odel and channel odel are described in Section 3. The hybrid precoding design is explained in Section 4, while Section 5 narrates siulation results. inally, e conclusions are ade in Section 6. 95

4 An optiized hybrid beaforing for illieter wave u-assive MIMO syste. MASSIVE MIMO AND MILLIMETER WAVE COMMUNICATIONS MIMO has becoe integral part of wireless standards as it can increase e reliability and roughput of e wireless counication systes [11] and one of e ost iportant key for current 4G wireless systes. Initially, MIMO was designed for e single-user, erefore, SU-MIMO becoe e integral part of long-ter evaluation and WiMAX [1]. Later it shifted fro single-user to ulti-user MIMO syste [1] where base station can counicate wi a group of users siultaneously wi ultiple antennas. To eet increasing data rates and traffic deand, currently, assive MIMO has becoe hot research issue where hundreds or ousands of antennas are installed at e base station to serve e group of users and iniize e interference. The Wave band is eerged as ost favorable band for assive MIMO..1. Benefits of MU- Massive MIMO MU-assive MIMO has attained a lot of attention for coing 5G wireless systes because of its high data rates for all e users and high reliability [13]. The ultiple antennas at are installed at transitter or receiver support ulti signal pas to get large roughput. By adopting MU-assive MIMO technique e syste capacity can be iproved by factor of 10 and spectral efficiency by factor of 0 when e nubers of antennas installed at transitter and receiver are about 56 [3]. Many ore benefits can be achieved which are as follows: i. Energy efficiency (EE) in odern cellular counications have growing attention worldwide. New research approach relative to EE called green cellular syste has driven e obile operators to eet e requireents of clean environent, decrease e cost and find solutions to iprove e EE for cellular obiles [14]. ii. In MU-Massive MIMO, we can reduce e nubers of aplifiers at uses a huge energy by replacing e traditional power aplifiers by a hundreds of low-cost power aplifiers [13], [15]. iii. Massive MIMO has a huge overflow of degree of freedo. Suppose we have 18 antennas at serve 8 users, en 100 degree of freedo reained unused [16].. Estiation of Channel Channel state inforation (CSI) is iportant for iproving e gain of ulti-antenna in MIMO syste. Therefore, CSI becoes ore coplicated for assive MIMO syste because e huge nubers of antenna at base station. In addition, assive MIMO eier applying DD or TDD requires a large nubers of pilots. In e presence of large nubers of users, e reuse of sae set of orogonal pilots in e neighboring cells can result e pilot containation in ulti-cell syste [17]. The pilot containation liits e perforance and decays e capacity of assive MIMO syste. When e nuber of antennas are increased at base station, e SINR at receiver side will increase too. Because e nuber of antennas at base station are equal to e nuber of pilot containations, e data signals for all users decreased. To solve is proble, e SIC is applied to itigate e interference between e pilot and data signals [18]. 96

5 An optiized hybrid beaforing for illieter wave u-assive MIMO syste.3. Counications in Millieter Waves Currently, all of e obile syste utilize spectru in e range of 300Mz to 30 Gz. The advanceent in electronic devices fabrication allows us to utilize e Wave spectru having large bandwid for fif generation (5G) wireless syste networks. Therefore, e 5G wireless syste ust test e 30 to 300 Gz spectru for e broadband counication syste. owever, e issues and challenges in e use of Wave spectru need to be addressed [19]. Availability of huge bandwid in Wave and provision of a hundreds and ousands of antennas at base station ake Wave and assive MIMO proising candidates for (5G) [1] [0]. The high-pa loss due to penetration and atospheric attenuation becoe e nightare for e Wave. Many researches have investigated it and advised to adopt sall cell syste networks such as fetocell, icrocell to overcoe e high-pa loss [1]. One of e ost suitable solutions to avoid e high atospheric attenuation transported by Wave is to utilize huge nubers of antennas to e base station. In past and current wireless systes onidirectional antenna array are used. The deployent of assive set of narrow-bea directional antennas can steer e signal direct to e ultiple users. There are a lot of antenna array techniques but e ost two fundaental techniques are e beaforing and spatial ultiplexing. In spatial ultiplexing, ultiple separately and independent encoded data streas can be transitted by using ultiple antenna eleents and e receivers have ability to reconstructing e original signal effectively [13, -5]. Spatial ultiplexing is ore effective in two cases: i. When e pa of prorogation is rich, it can support ultiple data streas transission. ii. At high channel SNR e streng of e original signal is not be affected when it splits in to ultiple data streas. In beaforing approach, e base station adapts e weight of for each array of antenna eleents to control e data streas direction [13]. In is way, e data strea targets e desired user and not interfere e oers. Beaforing is suitable in e conditions bellow: i. When e SNR of e channel is low. ii. When e power of e channel is liited. iii. Beaforing is critical for Wave to avoid e high-pa loss due to e sall waveleng. 3. SYSTEM MODEL We consider downlink ultiuser channel for hybrid and digital precoding, where one base station (BS) serves M N users in e cell. BS has t transit antennas and each user N equipped wi r receive antennas. Nuber of streas for each user are liited by N r N.Total nuber of R chains rf at transitter and total nuber of streas M in e downlink are liited by in( Nt, M Nr ) and in( Nrf, M Nr ) respectively. The R chains at receiver are assued equal to receiver antennas. The signal received at e user is given by: 97

6 An optiized hybrid beaforing for illieter wave u-assive MIMO syste M M M j j j j j y W P d W P d n Where n CN 0, is vector of i.i.d. noise and P (1) power allocated to e user. igure 1. Proposed MU-Massive MIMO Syste Syste Model for Digital Downlink Massive MIMO In digital downlink assive MIMO syste, e BS uses precoding atrix D D D W W1, W,..., W M D N t W at e transitter where is e atrix for D N r G user. A linear atrix ay be used at e d user to recover its sybols as: D D D M d G ( Wd n ) G ( Wi di n ) i () Where N N t r is e assive MIMO channel between e receiver. 3.. Syste Model for ybrid Downlink Massive MIMO ybrid beaforing cobines analog and digital precoders and decoders. If we expand e beaforing for ultiuser assive MIMO, e estiated signal at user can written as: d y y M y G R ( W i i d n ) i i (1) y N rf N N N N t rf r r W Where, R and are e atrices of baseband and analog precoding, and analog and digital decoding respectively. Every y y / eleent of R W got a constant odulus. or siplicity, we denote G and as W G and A for furer derivations. If we denote GR W (Cobining precoding and decoding of bo digital and analog), en e received SNR at user is: A( t, t) P t, M j A (, ) (, ) iu t t Q A j A t t Pj y G N r () 98

7 An optiized hybrid beaforing for illieter wave u-assive MIMO syste 3.3. Channel Model There are high free-space pa losses in Wave propagation environent at ust be characterized by an appropriate channel odel. We adopt cluster channel, described in [11]. This odel perfors e Wave channel atrix as: cl ray NN t r a a N N Ncl Nray r r t t 1 1 il r il i l il t il il (5) Where N is e nubers of clusters, N e nubers of rays for each cluster and cl e gain of l ray in e i prorogation channel is representing by il. 4. PROBLEM ORMULATION AND YBRID PRECODING DESIGN We want axiize e total surate of e syste, which is given as: arg ax log 1 s.t. M W RG tu, t M W M (6) The above proble is non-convex and it is difficult to obtain its optial solution. Therefore, we propose to solve an approxiation of (6) in order to find near-optial solution at can be practically ipleented. irst, we fix analog precoder and define effective channel atrix M eff (7) We take SVD of effective channel for digital precoder and decoder design. * eff eff eff eff U A V W V eff (8) U ray * eff and (9) We design e hybrid precoding in two steps, starting wi analog and digital decoders and en analog and digital precoder by e help of orogonal algori. As we can note fro equation (3) e atrices of hybrid are ore constrained an e atrices of digital beaforing, which eans digital beaforing achieves e best perforance. d Due to e advantages of block diagonalization, we eploy e as e received reference signal. We set e characteristics of proposed hybrid beaforing by finding e Euclidian distance between e received signal for digital and hybrid beaforing. d It can be acquired by estiating e ean square error between e D d and y at is given by [1]: D y D y d d d d (10) D We use e block diagonalization digital beaforing for total su rate axiization of ultiuser assive of downlink channel. The block diagonalization uses ree basic steps as follows: i- irstly, it eliinates e interference of oer users ii- Secondly it cancels e self-interference for each desired user. 99

8 An optiized hybrid beaforing for illieter wave u-assive MIMO syste iii- It optiizes e power of each sybol for axiizing e total su rate. or better clarification of e above steps we can define ese as follows: ˆ,,...,,... V h0 null ˆ X V Uˆ Z Vˆ h0 h h V h Where 0 N t unitary atrices and M (11) (1) null ˆ is e first factor of of, ˆ U h Nr V, ˆ h Z is e diagonal atrix of size. As we ention above e BD W beaforing algori can be carry out by putting and G of () as W ˆ Vh0Vh 0 Q G ˆ and Uh Q, e is denoted e atrix of power allocation for each user. Therefore, e relation between e input-output in (), can be rewritten as: ˆD ˆ ˆD d Z Q d Uhn di zi q d uˆ i hin (13) Where e ˆ D d i z q denoted e eleents of ˆ D d are and e diagonal eleent is represented by i i Z of Q uˆ and finally, e hi U denoted e row of ˆ h. The final steps of BD can be perfored by solving e total su rate axiization as: ax( q ) log 1 s.t. M ziqi i 1i1 M 1 i1 q i P ax P The eleents of axiu power at transitter ax (3) q is represented by e i. The ai is to solve is proble by utilize soe siple algoris such as water filling algori to reach e close-optial solution. R or e above equations we can recognize at, e BD beaforing Z Q R 0, and i i. Therefore, e in (10) can be written as: (4) ˆ ˆ ˆ G R W Z Q G R Wˆ Z Q G ˆ ˆ ˆ ˆ ˆ ˆ ˆ ˆ R W i G R W G R Uh G R Uh M i1, i Now we have to design analog precoder W, digital precoder ˆ R and analog decoder. The gap between digital and hybrid beaforing is constant for all e sybols. We use WSMSE algori to optiize e ain objective function i.e. e i sybol weight 1 of e z user is set to iqi, and e equation can be aeatically solved as: Wˆ R 1 1 in tr z Q z Q M 100

9 M tr W ˆ W ˆ Q ax s.t. 1 An optiized hybrid beaforing for illieter wave u-assive MIMO syste i, j M, Ri, j M (16) ax The M tr ˆ ˆ 1 W W Q is introduced to ake e sae total power for hybrid and digital beaforing. We furer optiize e proble in two steps: i. irst optiize e receiving atrix (digital decoder) and analog decoder for each user. ii. Then optiized analog and digital precoders for fixed analog and digital decoders irst Step R In e first step, we optiized jointly analog and digital G decoders for user, we written it as: h in W U, s. t. 1 W (5) W As we observed at and are joint function. Therefore, e constraint of e objective function is non-convex and optial solution of objective function is difficult to T find. To solve function in (17) we set e variable of f f where e null and rewrite e (17) as: h rf 0 in R G U, s. t. diag R R N G But even after soe aeatical anipulation e non-convex proble is still e sae. The better solution to overcoe non-convex proble is to use e orogonal atching pursuit algori as given below: (6) 101

10 An optiized hybrid beaforing for illieter wave u-assive MIMO syste Algori I. Orogonal algori to solve e (18) R [] Initialization: set R, R Uh. i 1: N or rf do R R i R R R R :, i argax diag 1 R h G R R R U R R end for U U R G h R G h 4.. Second Step G or a given R and in (18), e proble can be reforulated as: ax Wˆ s.t. G R W I Pax tr W W i, j 1 (7) We can notice for (15) at e correlation of analog precoder is uch high wi right singular value of. It is assued at each colun of analog precoder is derived fro one of e colun of f T NtNt f null, where wi noralized entries. ence e odulus value of analog precoder f is set to 1. By setting f and applying e sae technique as step1, e proble (19) can be rearrange as ax s.t. R G W I ax tr W W P diagww Nrf (0) Therefore, e solution to is proble can be found using e orogonal algori as used for (18) in step 1, as follows:, 10

11 An optiized hybrid beaforing for illieter wave u-assive MIMO syste Algori II. Orogonal atching for solving (0) Initialization: set i 1: N or rf do i R ax diag :, i ˆ RG 1 W ˆ ˆ ˆ ˆ R end for. W I Wˆ ˆ I Wˆ ˆ P ax Wˆ Wˆ RG and R, I It is entioned at e step 1 and step bo use fix e power at transitter side. 5. SIMULATION RESULTS N 18 MIMO This section presents e siulation results. We use, t N 3, r, where is Nt N is e nuber of antenna at BS and r is e nuber of antennas at each user s equipent. The nuber of users are M 4, antenna spacing d 0.5, e nuber of P av P SNR ax pas are L 16. The Pav is doinated by where and e total power is liited by Pax M. The channel paraeters and (θ t, (i) θ r, (i)) are selected randoly fro unifor distribution in [0, π]. Then e coputed su total rate is given as R M t 1 R i1 i, which eans e su total rate for each sybol is coputed by R i 10 log 1 i generated by 1000 iterations, R, where i is e achieved SINR of user i M. strea. Each plot is 5.1. Copression between ybrid and Digital Beaforing In is part, we copare e perforance of hybrid beaforing wi e digital beaforing. or siulation, we set e data streas to 1 and e nuber of radio N frequency chains rf to 16. The achievable rates for e digital and hybrid beaforing are shown in igure It can be observed at hybrid beaforing perforance 103

12 An optiized hybrid beaforing for illieter wave u-assive MIMO syste approaches e digital beaforig perforance fro low to ediu SNR region. In addition, it can be seen e sall gap perforance ay occur at high SNR region. igure. Digital and ybrid Beaforing Copression. N 5.. Joint Effects of and rf On ybrid and Digital Beaforing N In is section we study e joint effect of data streas and radio frequency chains rf on bo hybrid and digital beaforing perforance. We set SNR=5 db. igure 3 shows e perforance coparison of bo schees. It can be noted at by increasing e data streas, e achievable su rate of hybrid and digital beaforing will also P increase. The data streas are liited by e constant of axiu power ax M. By decreasing e data streas, e perforance gap between digital and hybrid can be N bridged. If we liit e nuber of radio frequency chain rf and e analog to digital converters (ADCs), e perforance gab between bo schees digital and hybrid will decrease by reducing e nuber of streas. igure 3. Joint Effects Of and N rf On Digital and ybrid Beaforing. 104

13 An optiized hybrid beaforing for illieter wave u-assive MIMO syste N 5.3. Effect of rf on hybrid beaforing In is part, e perforance of e hybrid beaforing is exained by varying e nuber of R chains. igure 4 presents of hybrid algori for deferent setting of N rf 8 : 0, at SNR = 5dB and 8. We can see at by increasing e nuber of R chains and ADCs, e perforance of e hybrid algori can be iproved significantly and e gap between e two schees becoes sall and alost constant when e nuber of R chains are ore an e twice of e data streas. igure 4. Effect of N rf on e ybrid Beaforing. 6. CONCLUSION We have designed a hybrid beaforing schee for downlink MU-assive MIMO syste. ybrid schee is indirectly designed by using WSMSE to axiize e total surate, whereas conventional block diagonalization is used for digital beaforing. We exaine e relation between two by considering different paraeters such as SNR, nuber of data sybols and e nuber of R chains and ADCs. ro e result we conclude at e perforance gap between hybrid and digital beaforing can be bridged by decreasing e nuber of data streas for a given nuber of ADCs and R chains. On e oer hand, when e nuber of data streas are fixed, an increase in R chains and ADC increases e total su rate of hybrid schee at we expect. The hybrid beaforing reaches e sae perforance of digital beaforing when SNR reaches fro e low to e high region. 105

14 An optiized hybrid beaforing for illieter wave u-assive MIMO syste 7. REERENCES [1] T. E. Bogale and L. B. Le, "Beaforing for Multiuser Massive MIMO Systes: Digital Versus ybrid Analog-Digital," Proceedings of e IEEE Global Counications Conference, Austin, USA, (014) Deceber 8-1. [] Z. Xinying, A.. Molisch, and K. Sun-Yuan, "Variable-Phase-Shift-Based R- Baseband Codesign for MIMO Antenna Selection," IEEE Transactions on Signal Processing., vol. 53, no. 11, (005), pp [3] Y. Gao, M. Khaliel, and T. Kaiser, "Wideband ybrid Analog-Digital Beaforing Massive MIMO Systes Based on Rotan Lens," Proceedings of e IEEE International Conference on Counication Systes (ICCS), Shenzhen, China, (016) Deceber [4] O. El Ayach, R. W. ea, S. Abu-Surra, S. Rajagopal, and Z. Pi, " Low Coplexity Precoding for Large Millieter Wave MIMO Systes," Proceedings of e IEEE International Conference on Counications (ICC), Ottawa, Canada, (01) June [5] E. Zhang and C. uang, "On Achieving Optial Rate of Digital Precoder by R- Baseband Codesign for MIMO Systes," Proceedings of e IEEE 80 Vehicular Technology Conference (VTC014-all), Vancouver, Canada, (014) Septeber [6] S. e, C. Qi, Y. Wu, and Y. uang, "Energy-Efficient Transceiver Design for ybrid Sub-Array Architecture MIMO Systes," IEEE Access., vol. 4, pp [7] A. Li and C. Masouros, "ybrid Analog-Digital Millieter-Wave MU-MIMO Transission wi Virtual Pa Selection," IEEE Counications Letters., vol. 1, no., (017), pp [8] S. Park, A. Alkhateeb, and R. W. ea, "Dynaic Subarrays for ybrid Precoding in Wideband Wave MIMO Systes," IEEE Transactions on Wireless Counications., vol. 16, no. 5, (017), pp [9] R. Rajashekar and L. anzo, "ybrid Beaforing in -Wave MIMO Systes aving a inite Input Alphabet," IEEE Transactions on Counications., vol. 64, no. 8, (016), pp [10] D. Vouyioukas, "A Survey on Beaforing Techniques for Wireless MIMO Relay Networks," International Journal of Antennas and Propagation., (013), pp.1-1. [11] Z. Pi and. Khan, "An Introduction to Millieter-Wave Mobile Broadband Systes," IEEE Counications Magazine., vol. 49, no. 6, (011), pp [1] M. K. Saii and T. S. Rappaport, "3-D Statistical Channel Model for Millieter- Wave Outdoor Mobile Broadband Counications," Proceedings of e IEEE International Conference on Counications (ICC), London, UK, (015) June 8-1. [13] J. Wu, Y. Zhang, M. Zukeran, and E. K. N. Yung, "Energy-Efficient Base-Stations Sleep-Mode Techniques in Green Cellular Networks: A Survey," IEEE Counications Surveys & Tutorials., vol. 17, no., (015), pp

15 An optiized hybrid beaforing for illieter wave u-assive MIMO syste [14] S. an, C. l. I, Z. Xu, and C. Rowell, "Large-Scale Antenna Systes Wi ybrid Analog and Digital Beaforing for Millieter Wave 5G," IEEE Counications Magazine, vol. 53, no.1, (015), pp [15] L. Dai, X. Gao, J. Quan, S. an, and I. C. L, "Near-Optial ybrid Analog and Digital Precoding for Downlink Wave Massive MIMO Systes," Proceedings of e IEEE International Conference on Counications (ICC), London, UK, (015) June 8-1. [16] A. Osseiran,. Boccardi, V. Braun, K. Kusue, P. Marsch, M. Maternia. O. Quese, M. Schellann,. Schotten,. Taoka and. Tullberg, "Scenarios for 5G Mobile and Wireless Counications: The Vision of e METIS Project," IEEE Counications Magazine., vol. 5, no. 5, (014), pp. 6-35, 014. [17] X. Yu, J. C. Shen, J. Zhang, and K. B. Letaief, "Alternating Miniization Algoris for ybrid Precoding in Millieter Wave MIMO Systes," IEEE Journal of Selected Topics in Signal Processing., vol. 10, no, 3, (016), pp [18] A. Alkhateeb, O. E. Ayach, G. Leus, and R. W. ea, "Channel Estiation and ybrid Precoding for Millieter Wave Cellular Systes," IEEE Journal of Selected Topics in Signal Processing., vol. 8, no, 5, (014), pp [19] O. E. Ayach, S. Rajagopal, S. Abu-Surra, Z. Pi, and R. W. ea, "Spatially Sparse Precoding in Millieter Wave MIMO Systes," IEEE Transactions on Wireless Counications., vol. 13, no, 3, (014), pp [0] M. N. Khan and M. Jail, Perforance Iproveent in Lifetie and Throughput of LEAC Protocol, Indian Journal of Science and Technology, vol. 9, no. 1, (016), pp [1] M. N. Khan and Jail, Maxiizing Throughput of ree Space Counication Systes Using Puncturing Technique, Arabian Journal for Science and Engineering, vol. 39, no. 1, (014), pp [] M. N. Khan,. K. asnain and M. Jail, Digital Signal Processing: A Breadfirst Approach, Rivers Publishers, Denark, (016), pp [3] J. G. Andrews, S. Buzzi, W. Choi, S. V. anly, A. Lozano, A. C. K. Soong, et al., "What Will 5G Be?," IEEE Journal on Selected Areas in Counications., vol. 3, no, 6, (014), pp [4] C. Li, Y. Ying, C. Xiaohui, and W. Weidong, "Multi-Stage Beaforing Codebook for 60 Gz WPAN," Proceedings of e 6 International ICST Conference on Counications and Networking in China (CINACOM), arbin, China, (011) August [5] M.. Alsharif and R. Nordin, "Evolution towards if Generation (5G) Wireless Networks: Current Trends and Challenges in e Deployent of Millietre Wave, Massive MIMO, and Sall Cells," Telecounication Systes., vol. 64, no. 4, (017), pp

16 An optiized hybrid beaforing for illieter wave u-assive MIMO syste 108