Introduction

One of the major requirement of 5G is to support the 1000 times more capacity per area as compared to that of LTE technology by keeping the same cost and energy dissipation per area. With the growing demand for wireless data traffic, the future cellular systems require higher area data throughput. By increasing the spectrum, number of BSs per area and spectral efficiency per cell, the capacity will be increased.

The spectral efficiency per cell can be increased by considering the Massive or large Multiple-Input Multiple-Output (MIMO) systems. When a large number of individually controllable antenna elements are utilized by a system at one side of communication link only, the system is said to be a massive MIMO system. Fig. 1 shows the usage of massive

MIMO at the BS.

Fig. 1. Multiuser and single user with massive MIMO base stations

The messages for severals users is multiplexed on the same time-frequency resource or the radiated signal is focused toward the intended receivers by massive MIMO network by exploiting spatial Degrees of Freedom (DoF) provided by number of antennas and hence, intra-cell and inter-cell interference can be minimized. Due to transmission of the same signal from multiple antenna points at a different phase shift, the radiated signals can be focused in a particular direction.

To increase the capacity of the system to several orders of magnitude, the three key approaches are suggested:

  • Ultra-dense networks (UDNs): 4G wireless cellular networks already adopt the network densification called small cell technology that can boost the capacity of the network.
  • Large quantities of new bandwidth: In order to achieve higher capacity, large amount of bandwidths are available due to migration towards higher frequencies.
  • High spectrum efficiency: There is a significant improvement of spectrum efficiency with the use of large number of antennas by massive MIMO as the available space resources can be extensively harnessed.

Hence, there is an increase in the capacity of the wireless system by using each of these approaches when compared to the current 4G LTE systems. A symbiotic convergence is shared by these approaches as there is a reduction in the physical size of the antenna array due to the very short wavelength of mmWave frequencies, short-range mmWave communications provides smaller cell sizes, while the severe path loss of mmWave signals can be overcome by the use of massive MIMO that provides large antenna gains. For 5G, the 1000-fold capacity increase can be achieved by using all these three approaches in a judicious way. In order to provide on demand coverage to the wide areas and localized small cell hotspots through mmWave technology, mmWave communications and massive MIMO are combined which is known as mmWave massive MIMO that will provide a platform with small cells of wireless network and high-speed data rates.

The need of accurate Channel State Information (CSI) at the transmitter side is one of the major challenge for the future 5G systems. The orthogonal pilot signals transmitted from each element of transmit antenna and the observed spatial channel feedback from the receiver to the transmitter side are used to obtain the CSI. There is a linear growth of the pilot signal overhead in terms of required CSI with the number of transmit antennas which is the major drawback of this approach. Utilization of the channel reciprocity also helps to obtain the CSI in TDD systems. There will be a degradation in the performance of a massive MIMO due to the delay between pilot transmission, channel estimation, channel feedback, beamformer calculation and the actual beamformed data transmission in the time varying channels. This delay can be mitigated with the use of channel prediction techniques.

Understanding the impact on the design of multi-cell multi-tier networks by the massive MIMO is another challenge. The impact of pilot contamination is one of the major problems. In multi-cell massive MU-MIMO systems, valuable insights are provided by the recent works into joint design of pilot and data channels.

MIMO in LTE

The major goal of designing the LTE with MIMO by 3GPP standard is to increase the capacity. Various MIMO technologies has been adopted by LTE. Upto four antennas are support by the DL transmission at the BS in LTE Release 8. The UL transmission supports only single antenna for transmission from the user. The option of antenna switching can also be provided that can switch up to two transmit antennas. UL can also support Multi-User MIMO (MU-MIMO). Moreover, enhanced MIMO technologies are provided by the Release 10 of LTE. The spatial multiplexing with up to eight independent spatial streams and enhanced MU-MIMO transmissions are supported with the implementation of a new codebook and feedback design. At the user side, there is a utilization of single user MIMO with up to four transmit antennas in the UL. By defining the new CSI reporting schemes, MIMO performance in DL was enhanced in Release 12 of LTE. Higher-order MIMO systems that provides upto up to sixty four antenna ports at the BS are used in LTE Release 13 so that high frequencies can be used appropriately.

Pilot design for massive MIMO

To recover the information that has been transmitted, there is a need to estimate the effect of the channel on the transmitted signal for conventional coherent receivers. The transmitted information can be recovered by the receiver by estimating the way the transmitted signal modified by the channel. There is an evaluation and proposal of large number of schemes such as blind, data-aided, and decision-directed non-blind techniques to estimate the channels. There must be an estimation of the channel’s effect on the transmitted signal for conventional coherent receivers so that the transmitted information can be recovered. The receiver can recover the transmitted signal only if it estimates how the channel modifies the transmitted signal. Due to the superior performance of the pilot signal-based data-aided techniques in the fast fading environment, these techniques are used which also provide cost efficiency and interoperability in commercial systems.

The use of pilot-based schemes is desirable when there is a large growth in the number of antennas at the BS due to which high-quality CSI has been provided for the detection of data at UL and for precoding at DL. Massive MIMO systems are dependent on the reciprocity of the channel and uplink pilots are employed to achieve accurate CSI at BSs. In case of non-reciprocal systems, the massive MIMO systems can only feasibly operate in TDD mode.

There has been contamination of the channel estimates by pilot reuse that is called Pilot Contamination (PiC) or pilot pollution. In massive MIMO systems, large number of channels has to be estimated, which is the biggest challenge for the future networks as availability of number of pilots is limited. The performance gains of non-cooperative MU-MIMO systems can be limited due to PiC. a saturation effect in the SINR can also be caused by PiC as there is a large increase in the number of antennas at the BSs.

There are two methods to combat the pilot contamination in massive MIMO systems i.e.

  • Open Loop Path loss Compensating (OLPC) based Pilot Power Control (PPC) in which efficient measures are provided against the detrimental effects of PiC.
  • The massive MIMO systems with coded random-access protocol in which new approach of mitigation of PiC has been brought.

Pilot contamination mitigation techniques

There is a reuse of pilot sequences in the nearby cells to reduce the pilot overhead that leads to PiC. The achievable performance of non-cooperative MU MIMO systems has been limited due to introduction of PiC.

There are large number of schemes to mitigate PiC in order to get the accurate PiC.

  • A precoding scheme: In this, the messages are linearly combined by each BS to transfer them to the users of different cells where the same pilot sequence is reused. This can resolve the PiC problem in BSs and tight pilot reuse is allowed.
  • Use of a Bayesian estimator: It mitigates PiC for spatially well-separated users. The Bayesian estimator can be implemented depending on the knowledge of the second-order statistics of the useful and crosstalk channels. Some overhead of estimating the covariance matrices and complexity of computation can be entailed by acquiring this knowledge due to which iterative filters are employed.
  • A low-complexity Bayesian channel estimator: It is also known as the polynomial expansion channel which is efficient in the presence of PiC.
  • Limited cooperation: It is based on the exchange of second-order channel statistics that uses Bayesian channel estimation. In this method, the same pilots are allocated to spatially well-separated users that can completely eliminate the PiC.

Resource allocation and transceiver algorithms for massive MIMO

The resources allocation among the users needs to be carefully designed so that the advantages of the massive MIMO can be exploited to offer the spatial multiplexing and array gain. The care must be taken for number of issues such as users and their count to schedule simultaneously, choice of antenna weights, design of the transceiver and scheduling etc.

In the first scheme, the main focus is on the designing of the the transceiver of a massive MIMO system in which the full exchange of CSI centrally between BSs can be avoided. At each BS, the optimal minimum power beamformers are obtained locally depending on limited backhaul information exchange between BSs.

The baptized interference clustering and user grouping is the second scheme in which large number of users in massive MIMO system are served by properly combination of user clustering, grouping and precoding.

MIMO Detection Schemes

MIMO detector has an objective of resolving the transmitted vector from the received signal. There are two classes of MIMO detectors:

  • Hard decision-based detector
  • Soft-decision-based detector

For hard decision, the decision of data symbols should be taken on the basis of the the detection with no extra estimation or information. Hence, for uncoded transmission, it is very useful.

In a soft-decision-based-detector, the log likelihood ratio (LLR) of each bit is calculated by using error correction coding scheme (ECE) and the bit correction is performed based on the estimation. As a result, there is an exchange of soft information between detector and decoding modules required by both iterative detection and decoding scheme. These detectors are suitable for subsequent iterative decoding and are called soft input soft output (SISO) detector.

Importance of Improving the Spectral Efficiency

The 5G technologies have a key goal of improving the area throughput by 100 and even 1000 times higher to keep up with the rapid traffic growth. The area throughput can be given as:

Area throughput = Bandwidth x Cell density x spectral efficiency

It is measured in bit/s/km^2. To obtain high area throughput, three main components can be improved:

  • Allocation of more bandwidth for 5G services
  • Addition of more cells with access points operating independently in order to densify the network
  • Improvement in the efficiency of the data transmissions

The cell densification and allocation of more bandwidth help to improve the area throughput in the previous network generations. In urban environments where the highest traffic demands are faced by the contemporary networks, there is a deployment of cellular networks with a few hundred meters inter-site distances and the availability of wireless local area networks (WLANs) has been increased. The good network coverage and service quality is provided by the frequency bands which are below 6 GHz and under short-range line-of-sight conditions, higher bands are used.

There is no improvement in the spectral efficiency (SE) in previous network generations. So, the future 5G networks have to improve this factor in order to achieve high area throughput. Massive multiple-input multiple-output (MIMO) technology provides the means to improve the SE of the future networks by one or two orders of magnitude.

Massive MIMO measurementsIntroduction

One of the major requirement of 5G is to support the 1000 times more capacity per area as compared to that of LTE technology by keeping the same cost and energy dissipation per area. With the growing demand for wireless data traffic, the future cellular systems require higher area data throughput. By increasing the spectrum, number of BSs per area and spectral efficiency per cell, the capacity will be increased.

The spectral efficiency per cell can be increased by considering the Massive or large Multiple-Input Multiple-Output (MIMO) systems. When a large number of individually controllable antenna elements are utilized by a system at one side of communication link only, the system is said to be a massive MIMO system. Fig. 1 shows the usage of massive

MIMO at the BS.

Fig. 1. Multiuser and single user with massive MIMO base stations

The messages for severals users is multiplexed on the same time-frequency resource or the radiated signal is focused toward the intended receivers by massive MIMO network by exploiting spatial Degrees of Freedom (DoF) provided by number of antennas and hence, intra-cell and inter-cell interference can be minimized. Due to transmission of the same signal from multiple antenna points at a different phase shift, the radiated signals can be focused in a particular direction.

To increase the capacity of the system to several orders of magnitude, the three key approaches are suggested:

  • Ultra-dense networks (UDNs): 4G wireless cellular networks already adopt the network densification called small cell technology that can boost the capacity of the network.
  • Large quantities of new bandwidth: In order to achieve higher capacity, large amount of bandwidths are available due to migration towards higher frequencies.
  • High spectrum efficiency: There is a significant improvement of spectrum efficiency with the use of large number of antennas by massive MIMO as the available space resources can be extensively harnessed.

Hence, there is an increase in the capacity of the wireless system by using each of these approaches when compared to the current 4G LTE systems. A symbiotic convergence is shared by these approaches as there is a reduction in the physical size of the antenna array due to the very short wavelength of mmWave frequencies, short-range mmWave communications provides smaller cell sizes, while the severe path loss of mmWave signals can be overcome by the use of massive MIMO that provides large antenna gains. For 5G, the 1000-fold capacity increase can be achieved by using all these three approaches in a judicious way. In order to provide on demand coverage to the wide areas and localized small cell hotspots through mmWave technology, mmWave communications and massive MIMO are combined which is known as mmWave massive MIMO that will provide a platform with small cells of wireless network and high-speed data rates.

The need of accurate Channel State Information (CSI) at the transmitter side is one of the major challenge for the future 5G systems. The orthogonal pilot signals transmitted from each element of transmit antenna and the observed spatial channel feedback from the receiver to the transmitter side are used to obtain the CSI. There is a linear growth of the pilot signal overhead in terms of required CSI with the number of transmit antennas which is the major drawback of this approach. Utilization of the channel reciprocity also helps to obtain the CSI in TDD systems. There will be a degradation in the performance of a massive MIMO due to the delay between pilot transmission, channel estimation, channel feedback, beamformer calculation and the actual beamformed data transmission in the time varying channels. This delay can be mitigated with the use of channel prediction techniques.

Understanding the impact on the design of multi-cell multi-tier networks by the massive MIMO is another challenge. The impact of pilot contamination is one of the major problems. In multi-cell massive MU-MIMO systems, valuable insights are provided by the recent works into joint design of pilot and data channels.

MIMO in LTE

The major goal of designing the LTE with MIMO by 3GPP standard is to increase the capacity. Various MIMO technologies has been adopted by LTE. Upto four antennas are support by the DL transmission at the BS in LTE Release 8. The UL transmission supports only single antenna for transmission from the user. The option of antenna switching can also be provided that can switch up to two transmit antennas. UL can also support Multi-User MIMO (MU-MIMO). Moreover, enhanced MIMO technologies are provided by the Release 10 of LTE. The spatial multiplexing with up to eight independent spatial streams and enhanced MU-MIMO transmissions are supported with the implementation of a new codebook and feedback design. At the user side, there is a utilization of single user MIMO with up to four transmit antennas in the UL. By defining the new CSI reporting schemes, MIMO performance in DL was enhanced in Release 12 of LTE. Higher-order MIMO systems that provides upto up to sixty four antenna ports at the BS are used in LTE Release 13 so that high frequencies can be used appropriately.

Pilot design for massive MIMO

To recover the information that has been transmitted, there is a need to estimate the effect of the channel on the transmitted signal for conventional coherent receivers. The transmitted information can be recovered by the receiver by estimating the way the transmitted signal modified by the channel. There is an evaluation and proposal of large number of schemes such as blind, data-aided, and decision-directed non-blind techniques to estimate the channels. There must be an estimation of the channel’s effect on the transmitted signal for conventional coherent receivers so that the transmitted information can be recovered. The receiver can recover the transmitted signal only if it estimates how the channel modifies the transmitted signal. Due to the superior performance of the pilot signal-based data-aided techniques in the fast fading environment, these techniques are used which also provide cost efficiency and interoperability in commercial systems.

The use of pilot-based schemes is desirable when there is a large growth in the number of antennas at the BS due to which high-quality CSI has been provided for the detection of data at UL and for precoding at DL. Massive MIMO systems are dependent on the reciprocity of the channel and uplink pilots are employed to achieve accurate CSI at BSs. In case of non-reciprocal systems, the massive MIMO systems can only feasibly operate in TDD mode.

There has been contamination of the channel estimates by pilot reuse that is called Pilot Contamination (PiC) or pilot pollution. In massive MIMO systems, large number of channels has to be estimated, which is the biggest challenge for the future networks as availability of number of pilots is limited. The performance gains of non-cooperative MU-MIMO systems can be limited due to PiC. a saturation effect in the SINR can also be caused by PiC as there is a large increase in the number of antennas at the BSs.

There are two methods to combat the pilot contamination in massive MIMO systems i.e.

  • Open Loop Path loss Compensating (OLPC) based Pilot Power Control (PPC) in which efficient measures are provided against the detrimental effects of PiC.
  • The massive MIMO systems with coded random-access protocol in which new approach of mitigation of PiC has been brought.

Pilot contamination mitigation techniques

There is a reuse of pilot sequences in the nearby cells to reduce the pilot overhead that leads to PiC. The achievable performance of non-cooperative MU MIMO systems has been limited due to introduction of PiC.

There are large number of schemes to mitigate PiC in order to get the accurate PiC.

  • A precoding scheme: In this, the messages are linearly combined by each BS to transfer them to the users of different cells where the same pilot sequence is reused. This can resolve the PiC problem in BSs and tight pilot reuse is allowed.
  • Use of a Bayesian estimator: It mitigates PiC for spatially well-separated users. The Bayesian estimator can be implemented depending on the knowledge of the second-order statistics of the useful and crosstalk channels. Some overhead of estimating the covariance matrices and complexity of computation can be entailed by acquiring this knowledge due to which iterative filters are employed.
  • A low-complexity Bayesian channel estimator: It is also known as the polynomial expansion channel which is efficient in the presence of PiC.
  • Limited cooperation: It is based on the exchange of second-order channel statistics that uses Bayesian channel estimation. In this method, the same pilots are allocated to spatially well-separated users that can completely eliminate the PiC.

Resource allocation and transceiver algorithms for massive MIMO

The resources allocation among the users needs to be carefully designed so that the advantages of the massive MIMO can be exploited to offer the spatial multiplexing and array gain. The care must be taken for number of issues such as users and their count to schedule simultaneously, choice of antenna weights, design of the transceiver and scheduling etc.

In the first scheme, the main focus is on the designing of the the transceiver of a massive MIMO system in which the full exchange of CSI centrally between BSs can be avoided. At each BS, the optimal minimum power beamformers are obtained locally depending on limited backhaul information exchange between BSs.

The baptized interference clustering and user grouping is the second scheme in which large number of users in massive MIMO system are served by properly combination of user clustering, grouping and precoding.

MIMO Detection Schemes

MIMO detector has an objective of resolving the transmitted vector from the received signal. There are two classes of MIMO detectors:

  • Hard decision-based detector
  • Soft-decision-based detector

For hard decision, the decision of data symbols should be taken on the basis of the the detection with no extra estimation or information. Hence, for uncoded transmission, it is very useful.

In a soft-decision-based-detector, the log likelihood ratio (LLR) of each bit is calculated by using error correction coding scheme (ECE) and the bit correction is performed based on the estimation. As a result, there is an exchange of soft information between detector and decoding modules required by both iterative detection and decoding scheme. These detectors are suitable for subsequent iterative decoding and are called soft input soft output (SISO) detector.

Importance of Improving the Spectral Efficiency

The 5G technologies have a key goal of improving the area throughput by 100 and even 1000 times higher to keep up with the rapid traffic growth. The area throughput can be given as:

Area throughput = Bandwidth x Cell density x spectral efficiency

It is measured in bit/s/km^2. To obtain high area throughput, three main components can be improved:

  • Allocation of more bandwidth for 5G services
  • Addition of more cells with access points operating independently in order to densify the network
  • Improvement in the efficiency of the data transmissions

The cell densification and allocation of more bandwidth help to improve the area throughput in the previous network generations. In urban environments where the highest traffic demands are faced by the contemporary networks, there is a deployment of cellular networks with a few hundred meters inter-site distances and the availability of wireless local area networks (WLANs) has been increased. The good network coverage and service quality is provided by the frequency bands which are below 6 GHz and under short-range line-of-sight conditions, higher bands are used.

There is no improvement in the spectral efficiency (SE) in previous network generations. So, the future 5G networks have to improve this factor in order to achieve high area throughput. Massive multiple-input multiple-output (MIMO) technology provides the means to improve the SE of the future networks by one or two orders of magnitude.


Join our massive mimo course