CHAPTER 3 REVIEW OF RESEARCH OUTCOMES IN MILLIMETER WAVE MIMO

CHAPTER 3 REVIEW OF RESEARCH OUTCOMES IN MILLIMETER WAVE MIMO 38

CHAPTER 3 REVIEW OF RESEARCH OUTCOMES IN MILLIMETER WAVE MIMO The advantage of MIMO is twofold link reliability and high data rate. These features when combined with MMW address the problem of penetration losses and multipath fading. Spatial diversity in transmitter and receiver is feasible in MMW as the size of the antenna is of the order of millimeter. This is not the case for low frequency where diversity is exploited only in fixed base station and not in the subscriber unit (SU). The reason being large sized multi-antenna in the SU affects portability of the device. The data rate of the order of gigabits per second is achieved owing to the unlicensed bandwidth and combining spatial multiplexing with MMW. This chapter deals with survey of reported results of MMW MIMO. The chapter is organized as follows: section 3.1 presents the MMW features and challenges, section 3.2 presents MMW MIMO, section 3.3 discusses the need for beamforming for 60 GHz system, section 3.4 presents the need for multibeam in MIMO, section 3.5 presents the 60 GHz channel sounder and section 3.6 presents the research outcomes in outdoor MMW MIMO. 3.1 Millimeter Wave features favoring use of MIMO The major features in brief and technical challenges in MMW that necessitate the need for MIMO is presented in the following section. The indoor channel in 57-64 GHz band experiences oxygen absorption at 60 GHz and also high free space path loss. The oxygen absorption limits the range to 10m indoor and 100m outdoor and this facilitates frequency reuse with interference free operation between the MMW systems. The propagation loss is overcome using high gain directional antenna. The directivity of an antenna scales inversely as the square of carrier wavelength. As a result of these directive antennae, the multipath environment is much sparser compared to lower carrier frequencies. Analysis of directional transmission and reception with electronically steerable beams using compact antenna arrays resulted in the compensation of propagation loss. Due to this, rich scattering environment was then reduced to sparse scattering leading to moderate antenna separation. This has led to 39

achieve spatial multiplexing in LOS. Capacity computations indicate data rate of gigabits is possible using channel bandwidths of the order of GHz (Torkildson, E. et al., 2010). Indoor channel model The channel model for WPAN is different from that of WLAN as the distance ranges are smaller, influence of human presence close to antenna device. These characteristics motivate the need for generalizing classical LOS MIMO model based on the assumption that antenna channel experience different small scale statistics and different channel gain. Hence MIMO channel model for WPAN is proposed and the simulation results on capacity, eigenvalue distribution and antenna correlation validated with measured data (Karedal, J. et al., 2010). License free huge bandwidth Making effective use of the unlicensed bandwidth, Ariza,A.P.G., proposed a method for high data rate polarimetric radio wave processing that finds use in short range communications and radar (Ariza, A.P.G., 2013). Tap-wise polarimetric filtering in delay domain enhances the 60 GHz link budget. This filtering enhances multipath from scatterers around the link and reduces multipath due to shadowing. In addition to this, MIMO polarimetry reduces heavy clutter for MMW radar. This help in capturing the complete information of the target and clutter (Ariza, A.P.G., 2013). Small Scale fading effects Small scale fading accounting for multipath effects dominates the signal propagation in indoor environment. MIMO small scale fading channel parameters in an underground gold mine analyzed by Mabrouk,I.B et al., in two scenarios, one in physically static environment and other in the presence of miners (Mabrouk, I.B. et al., 2013). As the propagation loss at 60 GHz is high, 2x2 MIMO antenna based on microstrip patch array was designed and fabricated. Using this experimental setup, various parameters like RMS delay spread, path loss, K-factor, channel correlation and capacity were analyzed. Table 3.1 shows the results (Mabrouk, I.B. et al., 2013). 40

Measurement RMS K-factor Path loss Correlation Capacity Scenario (ns) (db) exponent Transmit Receive (bits/s/hz) LOS 11.22 9.39 1.48 0.82 0.78 5.4 Human effect 24.86 7.90 2.4 0.75 0.58 7.44 Table 3.1 MIMO channel parameters analyzed in underground mine (Mabrouk, I.B. et al., 2013) Results in Table 3.1 indicate the presence of miners substantially affect both the received power and time dispersive parameters and hence should be considered while developing underground mine wireless networks. The overview of the MMW was discussed in length in chapter 2. The features that affect the channel capacity, link budget and channel modeling are summarized paving way for alternative solutions. As data rate of the order of gigabits with reliability is required in MMW system, researchers consider MIMO technique to meet the demand. 3.2 MMW MIMO This section presents the review of reported work in LOS and NLOS. The MMW research lab, University of California, Santa Barbara (UCSB) has proposed and developed 2x2 prototype hardware architecture that can be scaled to higher configurations. Similar such contributions are reviewed in this section. Capacity analysis for LOS Operating wireless links at optical speeds is demonstrated in 60 GHz band full duplex wireless Ethernet link. Speed of 1.25 Gbps is achieved in point to point link (Ohata, K. et al., 2003). This work stimulated wider interest of achieving data rate of the order of 10 to 40 Gbps. The MMW MIMO using MMW spectrum in the E-Band covering 71-95 GHz was considered as large spectrum was made available on semi-unlicensed basis by the FCC. Also this spectrum, avoided oxygen absorption at 60 GHz providing range of the order of kilometers. Owing to the fact of small wavelengths, high directive beams enabled spatial multiplexing with moderate antenna separation even in LOS link. Eigenmode beamforming using water filling and beam-steering with MMSE receiver using fixed constellation analyzed. The spatial multiplexing gain is robust to LOS blockage even with a small 41

transmit power per antenna element achieved using low cost CMOS process (Torkildson, E. et al., 2010). The authors extended the work using adaptive beamforming and spatial multiplexing to obtain capacity improvement. Adaptive beamforming using adaptive antenna array called subarray, enabled highly directive beam steerable over a larger angle improving the directivity gain in both transmitter and receiver. Subarrays were interpreted as single virtual elements in MIMO system with antenna spacing satisfying the Rayleigh criterion for realizing uniform array. This characteristic claimed the possibility of attaining high data rates of the order of 40 Gbps (Torkildson, E. et al., 2006). The concept extended to analyzing MMW in 57-64 GHz band for indoor environment. Authors (Torkildson E. et al., 2006) analyzed the performance fixing the range as 1km, which might not hold good if the range varies. The performance degradation is attributed to correlation between the spatial responses for different transmitters. Hence non-uniform arrays were proposed that reduced correlation over large set of ranges (Torkildson, E. et al., 2009). Capacity of short range systems using channel transfer function with angle of arrival (AoA) proposed. AoA data measured with the experimental setup used to calculate the 2x2 MIMO static channel capacity for varying antenna spacing of 1 mm to 20mm and transmit and receive distance of 27 cm (Liu, C. et al., 2010). Hardware prototype Hardware prototype demonstrating two channel spatial multiplexing operating at 1.2 Gbps in indoor office environment developed (Sheldon,C. et al., 2008). This was tested in indoor with 6.14 m link range and 12.4cm antenna element separations. Diffraction limited optics was used to establish multiple parallel data channels. This system can be scaled to longer link range, larger one and two dimensional arrays supporting data rate greater than 160 Gbps (Sheldon,C. et al., 2008). Scalable architecture proposed for four channel spatial multiplexing in LOS link using 1x4 linear arrays in transmitter and receiver. The system was tested in indoor office environment at 5m link range with antenna element spacing of 7.9cm. BER of less than 10-5 reported with signal to interference ratio of 11 db (Sheldon, C. et al., 2009). This replaces the manually tuned channel separation at Intermediate Frequency (IF) reported in (Sheldon, C. et al., 2008) with automatically tuned baseband channel separation. Enhancing the performance of 42

channel separation electronics in (Sheldon, C. et al., 2009) resulted in data rate of 2.4 Gbps (Sheldon, C. et al., 2010). Spatial degree of freedom Spatial degree of freedom is limited by the antenna array geometry, number of antennas and the scattering environment. Lower frequency systems rely on multipath for obtaining an uncorrelated channel leading to high multiplexing gain. But MMW systems, experience high propagation loss due to multipath that reduces the probability of uncorrelated channel. This is overcome by placing antenna with moderate separation with spatial degree of freedom preserved. Optimal array configuration has been analyzed such that LOS channel capacity is maximum with minimum number of antenna elements (Torkildson, E. et al., 2009). Capacity analysis for NLOS Minimum antenna spacing effects on capacity were analyzed. Antenna spacing of the order of the wavelength or fraction of wavelength was required for MIMO channels with both LOS and reflected path while few tens of wavelength was required for MIMO channel with only LOS. The reason attributed is LOS path suffers from blockages. Hence channel capacity of MIMO channel with LOS and reflected path was higher compared to MIMO channel with LOS (Lee, S.J. et al., 2010). Capacity with multiple elements at both antenna terminals at 60 GHz was studied for SIMO, MISO and MIMO configurations. The channel model utilized the geometric characteristics of the environment, angle of arrival, angle of departure of the propagation paths, the antenna elements and their spacing. Capacity of 4 bits/s/hz was estimated for transmit and receive distance of 15m (Arvanitis, A. et al., 2007). Study of improving the directivity and spatial multiplexing gain in the presence of multipath and LOS blockage led to the proposal of MMW MIMO architecture using array of subarrays (Torkildson, E. et al., 2011). Ray tracing model was used to simulate the indoor environment. Authors have compared the performance of the eigenmode transmission using waterfilling power allocation with beamsteering transmission and MMSE reception for a fixed signal constellation. The study provides insights of spatial variations of attainable capacity within a room. Promising multiplexing gain achieved by electronic beam steering for each subarray that makes the link robust to node placement and LOS blockage (Torkildson, E. et al., 2011). 43

3.3 Need for beamforming for 60 GHz system MMW radio suffers severe path loss, oxygen absorption and limited link budget. Methods to overcome the losses and improve link budget suggest using high gain directional antenna and beamforming technique. Blockage by metal objects deters the use of high gain directional antenna. Fixed and adaptive beamforming using MIMO antenna array form the best alternate in diffraction limited environment. The analysis with the objective of improving the link robustness, channel capacity, array gain, mixed beamforming techniques are discussed in length in the following sections. Beamforming analysis The influence of beamforming in LOS and NLOS scenario analyzed based on narrowband and wideband direction based beamformers for 5x5 and 7x7 planar array (Wyne, S. et al., 2011). Narrowband beamformer operating on 200 MHz bandwidth and wideband beamformer operating on 4 GHz full bandwidth were used to investigate in terms of improving the channel metrics i.e. delay spread, excess delay and SNR. The direction based beamformer was compared with various theoretical bounds. RMS delay spread and excess delay in beamformed LOS channel was around 1 ns and 0.5 ns. The array gain or SNR improvement was MN, with M and N representing the antenna elements in transmitter and receiver. The performance of direction based beamformer LOS channel was comparable to dominant eigenmode and statistical beamformer. Beamformed NLOS channel showed performance degradation relative to dominant eigenmode (Wyne, S. et al., 2011). The reported results are tabulated in Table 3.2. Channel metric Direction beamformed LOS Direction beamformed NLOS RMS Delay spread 1 ns 5 ns Excess Delay 0.5 ns 0.5 ns Array gain MN ( M 2 N) Table 3.2 Channel metrics using 5x5 planar antenna array (Wyne, S. et al., 2011). 44

Results in Table 3.2 indicate that the NLOS channels have few strong reflected components. Also influence of array size on beamforming was analyzed in addition to the channel metric. The 7x7 planar array had the same performance as the 5x5 array, thus setting limits for practically realizable array size. Weight vector for beamforming Beamforming vectors using iterative antenna training algorithm proposed (Xia, P. et al., 2008). This algorithm computes the Tx/Rx beamforming vector for both training and payload phase with reduced training overhead and is found efficient compared to antenna training. In antenna training entire channel co-efficients are estimated while in iterative algorithm two matrices one each in transmit and receive side are updated based on previous beamforming vectors (Xia, P. et al., 2008). Conventional baseband beamforming depends on channel matrix and thereupon determining the weight vectors with exact amplitude and phase. This method is quite impractical as it requires multiple RF chains that lead to high power consumption and also increases the complexity of the phase shifter and gain controller to cover a wider range of phases and amplitudes. Hence to minimize power consumption of RF devices, codebook based beamforming proposed (Wang, J. et al., 2009). The elements of the codebook are specified by one of the four phase shifts (0, 90, 180 and 270 degrees) without any amplitude adjustment. Multilevel training and antenna selection was considered in order to reduce the beamforming setup time (Lee, H-H. and Ko, Y-C. 2011). In every level transmit and receive antennas are selected based on inter-element spacing. This is followed with transmitting training sequences with different weight vectors from a pre-defined codebook. This helps the receiver to select the best transmit and receive weight vector to optimize the SNR. These vectors determine the codebook for the next level. This is computationally less complex relative to conventional codebook based beamforming that uses an exhaustive search transmitting all the training sequences before the best beam pair is selected (Lee, H-H. and Ko, Y-C. 2011). Efficient codebook based MIMO beamforming training scheme proposed for estimating antenna weight vectors (Zhou, L. and Ohashi, Y. 2012). Initially Discrete Fourier Transform (DFT) based codebook entries were used that had equal amplitude with low power consumption. The training algorithm then selects the best transmit weight vector and receive 45

weight vector as compared with multiple selection of weight vector in IEEE 802.11ad. Further refinement of weight vectors is performed by angular rotation around the best transmit and receive weight vector to improve the beam gain. The proposed work outperforms the IEEE 802.11ad standard in terms of efficient beamforming gain with less setup time (Zhou, L. and Ohashi, Y. 2012). Joint transmit and receive analog beamforming (ABF) proposed for systems with one scalar weight per antenna (Nsenga, J. et al., 2009). This is aimed at realizing low cost, low power MMW system. The scheme targets maximizing the average SNR at the input of the equalizer and analytically derive the close to optimal transmit/receive (Tx/Rx) scalar weights. The joint Tx/Rx ABF weights computed using channel state information (CSI). CSI is obtained from the inner product between all Tx/Rx channel impulse response (CIR) pairs. The BER performance of the joint Tx/Rx ABF outperforms the ABF scheme (Nsenga, J. et al., 2009). Low cost multi-antenna architectures with a lower number of analog front end chains than antenna elements proposed (Nsenga, J. et al., 2010). This enables mixed analog/digital beamformers that aim at maximizing the received average SNR. The performance found to be better than antenna selection techniques and digital beamforming (Nsenga, J. et al., 2010). Beamforming and Precoding attempted with large antenna arrays. Precoding in traditional MIMO systems is implemented in the baseband (Pi, Z. and Khan, F. 2011; Doan, C. et al., 2004). Baseband processing requires a dedicated radio frequency (RF) chain. This increases the cost of MMW RF hardware (Pi, Z. and Khan, F. 2011; Doan, C. et al., 2004). Thus precoding is shared between the analog and digital domain. Precoding in analog domain is done to reduce high cost of mixed -signal and RF chains. However precoding in the RF domain after upconversion is implemented using analog phase shifters (Zhang, X. et al., 2005; Nsenga, J. et al., 2010), that place a constant modulus constraint on the elements of the precoding matrix. Antenna selection, equal gain transmissions were proposed to address the problem of limited number of transmit/receive chains. These solutions do not account for large MMW systems, where transmit correlation is to be analyzed. Single user beamforming and precoding in MMW systems using large arrays is considered. The structure of MMW channel was exploited to design the precoder using sparsity constrained least square method. Principle of basis pursuit was used to develop the precoding algorithm that approximates unconstrained precoder using low dimensional basis representation. This results in low 46

complex RF hardware and the capacity of proposed algorithm was equal to that of MMW system using waterfilling method (Ayach, O.E. et al., 2012; Ayach, O.E. et al., 2012). MIMO postamble, used to reduce the complexity of implementation in MIMO beamformed system with SVD proposed (Tiraspolsky, S. et al., 2010). The complexity reduction is achieved by extracting CSI from the MIMO postambles that avoids the need for feedback. This leads to simple MIMO system architecture (Tiraspolsky, S. et al., 2010). Conventional beamforming with open loop transmit diversity studied and performance evaluated based on theoretical calculations. The hybrid scheme provides gain improvement over conventional beamforming with the same complexity (Pedersen, K.I. and Mogensen, P.E. 2001). Beamforming and closed loop transmit diversity in downlink proposed (Goransson, B. et al., 2004). The transmitted signal phase and amplitude on the diversity antennas is adjusted using feedback from each mobile user. With fixed antenna array, energy is radiated in fixed direction that may lead to interference to the mobile user other than the intended mobile user. To meet the Quality of Service (QoS), high power overcoming the interference is required. In an attempt to reduce the interference adaptive antenna array capable of steering the radiated energy to or from the mobile user is used (Goransson, B. et al., 2004). Near field beamforming addressing beampointing error is analyzed (Yan-ping, L. et al., 2012). The problem is formulated for uniform linear array. Convex optimization is applied to update the steering vector error of the beam, by maximizing the signal to interference and noise ratio subject to the constraint of minimum output variance of the source signal (Yanping, L. et al., 2012). Continuous Aperture Phased (CAP) MIMO based on the concept of beamspace MIMO that enables efficient access to the p communication modes of an n-dimensional MMW link is proposed, where p << n (Song, G-H. et al., 2013; Brady, J. et al., 2013). This is possible by exploiting the spatial dimension at MMW using high gain directional antenna proportional to n that leads to sparsity of propagation paths. This reduces the dimension of the communication subspace. Discrete lens array used to capture the beamspace channel matrix. The channel matrix is nearly diagonal indicating orthogonal Fourier spatial basis vectors used for beamforming. This is an indication of spatial multiplexing of data streams over p dimensional communication subspace. Thus using CAP MIMO enables near-optimal 47

communication with reduction in transceiver complexity. Mutual coupling is a matter of concern between closely spaced feed antennas which requires further investigation (Song, G- H. et al., 2013; Brady, J. et al., 2013). MIMO beamforming in multiuser environment Opportunistic beamforming in the transmitter that induces random beamforming in the network is reviewed. In order to have maximum opportunism, the system depends on perfect CSI. The authors Ozdemir, O. and Torlak, M., at University of Texas, Dallas have investigated the benefit of partial CSI at an opportunistic transmitter. The partial CSI information was basically the signal to noise ratio (SNR) relative to channel gain feedback by the user. The effect of relative SNR was then used to improve the throughput of the system (Ozdemir, O. and Torlak, M., 2008). The authors have extended the work by developing an optimum SNR quantization method that is performed at the mobile terminal. This development is later used to analyze the system throughput. Furthermore, the analysis indicates when the number of users in the system approach infinity, the performance of the opportunistic beamforming approaches to that of true beamforming (Ozdemir, O. and Torlak, M., 2010). Optimum zero forcing downlink beamforming analysed. The scheduling of the multiuser studied using greedy algorithm and water filling power allocation with average power constraint at the transmitter. The authors have derived joint probability density function for the scheduled users. The sum rate performance is evaluated and results indicate that the subchannel gains of the previously scheduled users remain unaffected. This is due to the cutoff value for water filling power allocation is suitably expressed (Ozyurt, S. and Torlak, M. 2012). The authors have extended the analysis in evaluating the zero forcing (ZF) Vertical Bell Labs Layered Space-Time (V-BLAST) algorithm using greedy decoder ordering in Rayleigh fading channels. The joint probability density function (pdf) of the squared layer gains assuming no error propagation is derived and is found applicable to any number of transmit and receive antennas. Based on this, joint distribution of the ordered gains from the joint distribution of the unordered gains is determined. This results in optimal power allocation using water filling algorithm avoiding extensive simulations to be carried out (Ozyurt, S. and Torlak, M. 2013). 48

3.4 Need for Multibeam in MIMO Multibeams are generated by antenna array. They form an integral part in MIMO systems. The capacity and performance of a system can be improved in two ways (i) At the baseband or at the intermediate frequency (IF) level by spatial multiplexing and diversity (ii) At the RF level by having antenna array and beamforming network (Godara, L.C., 1997). The diversity combining enhance the signal-to-noise ratio (SNR) of the received signal, while the beamforming network steers the beam to a direction where maximum gain is achieved (Yazd, M. H. and Faraji-Dana, R., 2005; Bellofiore, S. et al., 2002). Combining the above two methods may result in increase in both capacity and performance. This also leads to reduction in interference in multiuser environment (Tsunekawa, K. 2005). The multibeam generated by multi-antenna system is shown in Figure 3.1 (Tsunekawa, K. 2005). In Figure 3.1, the antenna element and multi-antenna system are shown. Figure 3.1 Multi-antenna system with baseband, IF and RF subsystems (Tsunekawa, K. 2005) 49

The baseband unit, RF and antenna element together form the multi-antenna system in Figure 3.1. High gain in antenna element is achieved by adding a parasitic element or integrating an RF circuit. The radiation pattern is controlled by baseband signal processing and hence pattern of each wireless channel can be controlled by each baseband channel. Thus, multiantenna systems are generally called smart or active antennas (Alexiou, A. and Haardt, M. 2004). Multibeam selection scheme, selecting the best subset of all beams to maximize capacity using orthonormal random beamforming with partial CSI is proposed (Zhang, K. and Niu, Z. 2006). The scheme is compared with dirty paper coding that uses full CSI and is found to have performance improvement under low SNR (Zhang, K. and Niu, Z. 2006). Diversity and Multiplexing gain for multi-user MIMO using orthogonal multiple random beams proposed (Oho, D.-C. and Lee, Y.-H. 2005). Orthogonal beams enable users to experience multiple channels at the same time realizing multi-user diversity. Also multibeams increase the multiplexing gain. Both MISO and MIMO configuration benefit from the scheme (Oho, D.-C. and Lee, Y.-H. 2005). Multibeam assisted MIMO (MBA- MIMO) proposed as an alternative for fixed beamfomer that suffers from inter-beam interference proposed (Papadogiannis, A. and Burr, A.G. 2011). MBA-MIMO is applied in the beam domain to mitigate interference and enable the provision of required capacity density (Papadogiannis, A. and Burr, A.G. 2011). 3.5 60 GHz channel sounder Measurement of capacity and double directional parameters with 16 x16 MIMO systems was performed in LOS, LOS obstructed and NLOS (Ranvier, S. et al., 2005; Foschini, J.G. 1996). Direction of Arrival (DOA) and Direction of Departure (DOD) were calculated from LOS measurements and channel capacity was calculated from NLOS. MIMO capacity obtained in NLOS is less compared to i.i.d Rayleigh channel (Ranvier, S. et al., 2005; Foschini, J.G. 1996). 2x2 MIMO capacity measured using 4-port Agilent Network Analyzer. Open-ended waveguide antennas with different orientations are used in the study. The study confirms the achievement of enhanced data rate assessed using the Shannon capacity theorem (Ahmadi- Shokouh, J. et al., 2012). 50

3.6 Outdoor MMW MIMO Outdoor deployment of MMW was analyzed, investigating the robustness of the link using ray tracing channel model (Zhang, H. et al., 2010). This study indicated the sensitivity of the received power to small variations in propagation geometry, which has led the analysis to employ multiple antennas. Robust performance was reported based on antenna separation, transmit precoding and space time coding (Zhang, H. et al., 2010). A two channel hardware prototype operating at 1.2 Gbps over a link range of 41m highlighted the potential of spatial multiplexing in LOS link (Sheldon, C. et al., 2008). The prototype achieved a BER of less than 2x10-6 and data rate of 600 Mbps per channel. The antenna spacing as a function of wavelength and link range determined in interference limited propagation (Sheldon, C. et al., 2008). Multilevel beamforming in outdoor MMW communication systems using beam searching algorithm and codebook design proposed (Hur, S. et al., 2011). The codebook controls the beamwidth and steering direction. The study is used in the design of MMW backhaul systems supporting picocell data traffic. The codebook is designed based on adaptive beamwidth beamforming technique developed on subarray using squinting angles. As beamwidth is inversely proportional to the array size (Trees, H.L.V. 2002), sharp beam is obtained using the entire array, while this beam can be broadened using subarraying and squinting. Each level of the codebook uses a different subarray size resulting in varied beamwidth. Directional gain issue arising out of squinting and subarraying is avoided using spectral windowing. The performance of the exhaustive search technique when compared with proposed scheme offers better beamforming gain at the cost of increased complexity. But the proposed joint search method displays larger gain compared to one-side search with the same complexity (Hur, S. et al., 2011). 51

Summary of the comparative description of various reported results of MMW MIMO worldwide. a) Ariza,A.P.G., et al., (2013), IEEE Transactions on Antennas and Propagation. Multipath effects Polarimetric radio wave processing. Enhanced multipath from resulting from Polarimetry is used to analyze the scatterers around the link scatterer and characteristics of electromagnetic waves. and reduces multipath due to shadowing shadowing. In addition to Tap-wise polarimetric filtering in delay this, MIMO polarimetry is domain enhancing the 60 GHz link budget. proposed to reduce heavy clutter for MMW radar. (a) b) Mabrouk,I.B. et al., (2013), IEEE Transactions on Antennas and Propagation MIMO channel 2x2 MIMO antenna based microstrip parameters in an patch array designed. underground gold mine (RMS delay spread, K-factor, path loss exponent, Tx/Rx correlation, capacity) Parameters were studied in two scenarios: one in physically static environment and other in the presence of miners. Table 3.1 lists the measured data for the above two cases. (b) c) Song, G-H et al., (2013),International Conference on Acoustics, Speech and Signal Processing d) Brady, J. et al., (2013), IEEE Transactions on Antennas and Propagation (c) The presence of miners substantially affects both received power and time dispersion parameters and the effects to be considered when developing underground mine wireless networks in the unlicensed 60-GHz band. Channel matrix Continuous Aperture Phased (CAP) Channel matrix is nearly MIMO based on the concept of diagonal indicating beamspace MIMO that enables efficient access to the p communication modes of an n-dimensional MMW link, where p orthogonal Fourier spatial basis vectors used for beamforming << n. Exploiting the spatial dimension at MMW using high gain directional antenna proportional to n that leads to sparsity of propagation paths. This reduces the dimension of the communication subspace. 52

e) Ozyurt, S. and Torlak, M. (2013), IEEE Transactions on Wireless Communications Sum rate and Downlink Beamforming in multiuser Subchannel gain environment Zero Forcing V-BLAST using greedy decoder ordering in Rayleigh fading channel. (d) f) Ayach, O.E. et al., (2012) IEEE International Conference on Communications. g) Ayach,O.E. et al., (2012), IEEE International Workshop on Signal Processing Advances in Wireless Communications. h) Liu. C. et al., (2010), IEEE Transactions on Wireless Communications. (e) (f) The joint probability density function assuming no error propagation derived. This results in optimal power allocation using water filling algorithm. Channel capacity and RF hardware complexity Beamforming and precoding in MMW systems with large arrays. Principle of basis pursuit was used to develop the precoding algorithm that approximates unconstrained precoder using low dimensional basis representation. Capacity is equal to that achieved using water filling algorithm. Channel Capacity Angle-of-Arrival information used to LOS predict the static channel capacity. signal contributes An Anritsu 37397 Vector Network Analyzer (VNA) used to measure the channel transfer function. significantly to the spatial MIMO channel capacity for short distance applications. i) Ozdemir, O. and Torlak, M. (2010), IEEE Transactions on Wireless Communications SNR and Optimum SNR quantization performed System throughput is Throughput at the mobile terminal improved. The performance of opportunistic Opportunistic beamforming in the beamforming approaches transmitter for flat and slow fading true beamforming when channel. number of users in the system approach infinity. (g) 53

j) Sheldon, C. et al., (2009), IEEE- MTTS International Microwave Symposium. Four channel Increased communication spatial multiplexing Scalable system architecture is proposed and demonstrated for spatial multiplexing over MMW line-of-sight communication. The system was tested in indoor office environment at 5m link range with antenna element spacing of 7.9cm k) Wang, J. et al., (2009), IEEE Vehicular Technology Conference. l) Papadogiannis, A and Burr, A.G. (2011), Future Network & Mobile Summit. (h) Linear detection technique such as Zero forcing (ZF), Minimum Mean Square Error (MMSE) are applied in beam domain instead of the antenna domain. (j) link data capacity without increased channel bandwidth. BER of less than 10-5 reported with signal to interference ratio of 11 db Power Codebook based beamforming used. Power minimization Minimization of achieved compared to RF devices. The elements of the codebook are conventional baseband specified by one of the four phase shifts beamforming. (0, 90, 180 and 270 degrees) without any amplitude adjustment. (i) Interference and MIMO techniques are Capacity density for future Capacity applied in the beam domain. systems beyond 4G. m) Oho,D.-C., Lee,Y.-H., (2005), IEEE Vehicular Technology Conference. Diversity and Orthogonal multiple random beams in Multiplexing Gain MIMO. Orthogonal beams enable users to experience multiple channels at the same time realizing multi-user diversity. Also multibeams increase the multiplexing gain. (k) Applicable to multi-input single-output (MISO) schemes as well as MIMO schemes, enabling the use of flexible antenna structures in the receiver 54

n) Zhang,K. and Niu,Z. (2006), IEEE International Conference on Communications Computational Complexity Multibeam selection scheme, selecting the best subset of all beams to maximize capacity using orthonormal random beamforming with partial CSI is proposed. Orthonormal random beamforming which only requires partial CSI feedback at the transmitter. This technique results in less computational complexity than dirty paper coding. (l) Efficient under low SNR o) Tsunekawa, K. (2005), NTT Technical Review. Antenna Gain High-efficiency antenna created by integrating it with a monolithic microwave integrated circuit (MMIC) in a millimeter-wave frequency-band system,a multi-antenna system for the MIMO technique that enables multiple data transmission for land mobile communication systems. High-efficiency antenna can be made using an active integrated antenna, which integrates an MMIC with the antenna element. The baseband unit, RF and antenna element together form the multi-antenna system. High gain in antenna element is achieved by adding a parasitic element or integrating an RF circuit. (m) Table 3.3 (a m) Survey of reported results in Millimeter Wave MIMO 55

3.7 General observations based on present Research work The chapter has extracted the contributions of researchers in MMW MIMO with the following concluding points in addition to the salient features presented in Table 3.3 and spatial multiplexing in LOS possible by high gain directional antenna leads to enhanced data rate. The losses at 60 GHz, small wavelength, oxygen absorption, unlicensed bandwidth, poor link budget and LOS blockage motivate the use of antenna beamforimg. Beamformer reduces the interference and ensures increased received signal power. The phase and amplitude of the weight vectors of the beamformer are fixed or can be adapted based on the CSI fedback to the transmitter. Implementation of fixed antenna array is preferred compared to adaptive array as this requires complex phase control network. Solutions optimizing the cost of RF chain have been reported. Dedicated RF chain per antenna increases the cost and complexity of RF components, hence antenna selection and subarray technique are reported. Methods of diagonalizing the channel matrix are analyzed with an objective of reducing the computational complexity. Continuous Aperture Phased (CAP) MIMO based on the concept of beamspace MIMO enables efficient access to the p communication modes of an n-dimensional MMW link is reported, where p << n. Conventional beamforming with open loop transmit diversity studied and performance evaluated based on theoretical calculations. The hybrid scheme provides gain improvement over conventional beamforming with the same complexity. Experimental studies on channel parameters have considered Ray Tracing based on geometric optics, Saleh Valenzuela (SV) a cluster based model for estimating the channel parameters such as the RMS delay spread, Angular spread, Rician factor. These parameters have been studied for indoor environment such as conference room, desktop, hospital and underground gold mine. Various studies indicate the indoor system using MMW technology possess time dispersive and doppler sparse environment. Also using directional antenna causes sparsity of propagation paths that enables compressive sensing technique for channel estimation. Multilevel beamforming in outdoor MMW communication systems using beam searching algorithm and codebook design proposed. The codebook controls the beamwidth and steering direction. The study is used in the design of MMW backhaul systems supporting picocell data traffic. 56

Multibeam MIMO proposed with an objective of reducing beam interference most likely in fixed beamformer using MBA-MIMO. Orthogonal multiple random beams in MIMO enables improvement in diversity and multiplexing gain. Also antenna gain and efficiency improvement using multibeam MIMO reported. MMW-MIMO needs more refined system components in terms of antenna systems, channel parameters etc. Survey report presents wide scope of research as no complete system design with fully satisfied results reported. The proposed research work, we have carried out in this tenure is not found in any of the reports that proves it may be unique in nature. Future 5G systems, propose to use MMW as the last mile solution that is cost-effective alternative to the existing standards. Samsung Electronics announced that it has successfully developed the world s first adaptive array transceiver technology operating in the millimeter wave Ka bands for cellular communications. The new technology sits at the core of 5G mobile communication system and will provide data transmission up to several hundred times faster than current 4G networks. Samsung plans to accelerate the research & development of 5G mobile communication technologies, including adaptive array transceiver at the MMW bands, to commercialize those technologies by 2020 (http://global.samsungtomorrow.com/). Based on this survey, it is very clear that MMW MIMO still needs to be refined to achieve its full potential. The main challenge of integrating the RF front end system with baseband signal processing system is still a formidable task due to the complexity in the antenna design and associated systems for mobile devices. The free space path loss and atmospheric loss added constraints to be tackled using beamforming concept. Omnidirectional antenna systems are not a good solution in terms of power optimization. Since indoor communication environments are considered, the rain attenuation need not be accounted. In summary, MMW MIMO is a wonderful pair of technology that will tend to have extraordinary potential for establishing wireless networks with gigabit capacity. Thus summarizing on the research outcomes, MIMO for MMW was analyzed for LOS and NLOS conditions, need for beamforming and multibeam MIMO. Ray tracing channel model, empirical model was used to model the indoor channel and study the system under test. As this model is site-specific and throws constraints in evaluating system performance, statistical channel models have to be considered. Also both open and closed loop beamforming techniques proved to yield better results throws scope for extending the idea to multibeam 57

MIMO. The challenges brought out in the reported results lays path for exploring the cluster based statistical channel models in LOS supporting multi-antenna configuration that is discussed in detail in chapter 4. 58