Advanced Search >>

« July 2010 | Main | September 2010 »

1 posts from August 2010

08/10/2010

The Masters of MIMO


August 2010

Mom

This month, we asked our contributing experts to "The Masters of MIMO" August cover story to respond to audience questions from our last webinar on MIMO. Below are the individual replies from Nicolas Gross, Alessandro Scannavini and Meryam Abou el Anouar (SATIMO), Derek Skousen (MI Technologies), Madhusudhan Gurumurthy (Spirent Communications) and Moray Rumney (Agiloent Technologies). Do these experts agree on their responses?

To comment on their responses or ask the Experts from SATIMO, MI Technologies, Spirent or Agilent a follow-up question, use the comment link at the bottom of the entry.

Q: Many of these MIMO techniques require a stationary environment after channel estimation training. How can a mobile system whose environment is always changing be accommodated ?

 

Derek Skousen – Product Marketing Analyst:

 

 

 

MI Technologies

Nicolas GROSS – Applications Director , Mr. Alessandro SCANNAVINI - Field Application Engineer and Meryam ABOU EL ANOUAR - R&D Antenna Engineer:

 

SATIMO

There are several different approaches to channel estimation training: instantaneous, statistical, pilot tone based, integrated, etc. Each of these trades off things like speed, processing power, complexity, bandwidth usage, or cost as they try to accommodate some of the channel variations.

    Even though a mobile channel is constantly changing, many changes are slow enough that a stationary environment is not a bad assumption over the short time periods between estimations. Other variables change too fast for almost any channel estimation technique and other portions of the system must be designed to handle them. In either case, practical designs will only catch a certain fraction of their potential – but in today’s stressed wireless systems, every fraction can be worth it.

    It's important to remember that MIMO implementations are much bigger than just the DSP subsystem (where features like channel estimation are usually implemented). Board, chassis and antenna designs create the interface between the device and the channel. Failure to optimize this hardware interface can handicap subsystem designers to the point that no amount of design brilliance can eek out the incremental gains MIMO advertises

 

Spatial channel models are typically used for modeling MIMO scenarios since they include the necessary spatial and temporal characteristics of a typical wide-band cellular channel. By using SFE (Spatial Fading Emulation) technique, channel parameters would not change dynamically. It is like testing the phone by considering a snapshot of the scenario the phone will experience when working in the field. Parameters like delay spread, and Doppler spread will be based on the standard channel models used for.

 

Moray Rumney

Lead technologist

 

Agilent Technologies

 

One of the benefits of LTE over previous systems is the shorter sub-frame which means the response time of the system to UE-generated channel state information (CSI) is faster than before. This means LTE can adapt more quickly to the changing environment in around 4 ms with retransmission of lost packets in 8 ms. This compares favorably with the 12ms retransmission latency of HSPA. However, the value of this faster response time is only as good as the accuracy of the CSI. Current conformance test procedures are primarily open loop and do not fully stress the instantaneous accuracy of the CSI.

 

 

Q: For Real-world, over-the-air testing what approach works best for duplicating the wide range of channel fading and multipath conditions that will provide an adequate evaluation?

 

Derek Skousen – Product Marketing Analyst:

 

 

 

MI Technologies

Nicolas GROSS – Applications Director , Mr. Alessandro SCANNAVINI – Field Application Engineer and Meryam ABOU EL ANOUAR – R&D Antenna Engineer:

 

SATIMO

The challenge for real-world OTA testing is in creating a repeatable reference condition that multiple devices can be tested against.   In the past, this was not as difficult because the device designs didn’t have to find extra system gains from the channel.  This is no longer the case with MIMO implementations.  The value of MIMO functionality is inseparable from varying channel impairment conditions like multipath.  Today there are two approaches for creating changing reference channel environments to simulate the great diversity of multipath signals:

 

1) An anechoic chamber with 4 to 16 antennas surrounding the device.  In this approach, a channel emulation signal is fed into these antennas to create an RF environment that simulates a channel.  In addition to being a complex setup, this technique has at least three major limitations: first, the ring of antennas only creates a 2D plane of simulated space; second, the limited number of calibrated antennas restricts angular variations; and third, emulated channels and their variables need to be restricted to keep uncertainty and test times reasonable and to maintain commonality across labs. Thus the anechoic chamber approach provides a very limited subset of the wide range of channel fading and multipath with a significant penalty in the potential issues that arise in a complex test set up.

 

2) A reverberation chamber, which is a highly reflective cavity that creates standing waves, with mechanical stirrers that create and expose the DUT to a full distribution of the highly variable multipath 3D propagation environment. In this approach, statistical analysis of many simple measurements quickly summarizes device effectiveness.  Thus the reverberation chamber approach provides a more general and complex reference channel environment in a relatively simple and fast test set up and has been demonstrated to provide highly repeatable measurements.

 

A Spatial Fading Emulator approach, combining an anechoic chamber, several probes and a channel emulator, can emulate any specific environment including a variety of standardized channel models currently used for UE conformance testing based on 3GPP standard. With such a setup, end to end testing of the wireless device, including its antenna configurations, receiver, digital signal processing and software, can be performed in a controlled realistic environment.

Madhusudhan Gurumurthy – Senior Applications Specialist

 

Spirent Communications

Moray Rumney

Lead technologist

 

Agilent Technologies

Investigations in this area are continuing and are being monitored very carefully by the 3GPP. In our experience so far the “reverb chamber” approach seems to reduce the antenna behavior into an average gain value, which tends to overestimate the performance and miss most of the corner cases. In our estimation these corner cases offer the most value and are the best indicators of post-deployment performance. We’ve seen very good emulation when using an anechoic chamber with multiple RF channel emulators to enable a number of TX antenna probes.  This technique allows throughput to be evaluated as a function of the device orientation, which leads to a more precise evaluation of the figure of merit. 

 

This is a highly relevant question whose answer is far from being agreed. The articles in this issue that describe the two main methods (anechoic and reverb) for creating a complex spatial environment will provide more detail than can be repeated here. The debate over exactly which channel model is necessary will continue for some time. It is important to state a couple of points; firstly that it is desirable but not necessary for the test environment to exactly match a real world environment in order to make useful measurements and secondly, the absolute accuracy of the environment (not just its repeatability) is of paramount importance, particularly when measuring figures of merit related to throughput. One proposal from Agilent to resolve this debate is to define reference “good” and “bad” devices then allow candidate test methods to demonstrate repeatable measurements that can consistently tell the difference between good and bad devices. Until we can reach this point the debate is likely to continue.

 

Q: In a system that is MIMO capable, is there a possible performance advantage when only one antenna is active compared to a system that has only single antenna capability?

 

Derek Skousen (MI Technologies) replies:

Typically, no. In fact the presence of the other antenna(s), even if not active, can degrade the performance of the active antenna compared to a SISO device.

 

Q: What is the effect of the mutual coupling on the performance of the MIMO?

 

Derek Skousen – Product Marketing Analyst:

 

 

 

MI Technologies

Nicolas GROSS – Applications Director , Mr. Alessandro SCANNAVINI - Field Application Engineer and Meryam ABOU EL ANOUAR - R&D Antenna Engineer:

 

SATIMO

Mutual coupling reduces antenna efficiency.  Since small MIMO devices are subject to these effects, the gains in multipath performance need to be able to offset the losses due to mutual coupling. By careful antenna design, this mutual coupling effect can be minimized allowing for measurable MIMO advantages. I should emphasize that the antenna design in this case refers to an entire system design at the device level since chassis, boards, materials, paints, nearby components, etc. can significantly affect antenna efficiencies.

 

It increases the correlation between the antennas signals, then the more coupling, the less advantage can be achieved from the presence of multi signals.

Moray Rumney

Lead technologist

 

Agilent Technologies

 

I’ll answer this question using an acoustic analogy: consider how the value of a stereo audio signal degrades as the coupling between left and right channels increases. The best way to experience stereo is with closed headphones where there is virtually no coupling. The worst way to experience stereo is in a reverberant room with the speakers too close together. In MIMO, coupled antennas and antenna gain imbalance are probably the worst things that can happen for system performance since this directly increases the required SINR for the system to enable spatial multiplexing to operate.

 

 

Q: What are some of the leading factors that influence multiple antenna correlation, isolation and efficiency?

 

Derek Skousen – Product Marketing Analyst:

 

 

 

MI Technologies

Nicolas GROSS – Applications Director , Mr. Alessandro SCANNAVINI - Field Application Engineer and Meryam ABOU EL ANOUAR - R&D Antenna Engineer:

 

SATIMO

Proximity and alignment are the leading factors for antenna correlation and isolation, while size is clearly the leading factor for efficiency. For all three, the user device itself creates ground plane variations or resonator effects that will influence these measurements. This underscores the need for rapid testing of correlation, isolation and efficiency to allow for dynamic optimization of an antenna design in the device.

We do recognize element locations, spacing between elements, and radiation characteristics of individual elements (radiation pattern, polarization) as the leading factors. These parameters must be jointly considered in order to achieve optimum performance trade off for multiple antenna terminals.

Madhusudhan Gurumurthy – Senior Applications Specialist

 

Spirent Communications

Moray Rumney

Lead technologist

 

Agilent Technologies

The correlation observed in a MIMO channel depends on a variety of factors such as device geometry (orientation, antenna pattern, antenna spacing) and the scattering environment (angle spread, power azimuth spectrum). Many of these factors constantly vary due to motion and dynamic environment conditions, resulting in varying correlations. In general, environments that are rich in scattering produce large angle spreads and result in low spatial correlation. Uncluttered areas tend to produce less multi-path resulting in narrow angle spread and higher spatial  correlation.  Interestingly, data collected in multiple environments tend to show that there is almost always a narrow angle spread. The common assumption of a uniform distribution of signals is virtually never observed.

 

The usual suspects of antenna spacing, polarization and size are generally understood and will dominate the free space performance of any MIMO antenna system. What is less well understood is the impact of realistic loading caused by the proximity of head, hand or laptop in the case of a dongle. There has been some fascinating research recently presented by Molex to COST2100 which showed that it is very hard, particularly at low frequencies to design for good de-correlation. Devices that have low correlation in free space may be detuned by head and hand loading and ironically, devices that have high correlation tend to get better when loaded. such that all loaded devices end up having similar medium performance. The conclusion from this study is that antenna gain imbalance has a larger impact on performance than practical correlation but even more important is the need to design the antennas such that the impact of head and hand loading is minimized. A poorly placed antenna system can have serious consequences on performance.

 

 

Q: What is multipath fading and what are some of the environmental concerns that can impact the signal path profile propagation characteristics?

 

Derek Skousen – Product Marketing Analyst:

 

 

 

MI Technologies

Nicolas GROSS – Applications Director , Mr. Alessandro SCANNAVINI - Field Application Engineer and Meryam ABOU EL ANOUAR - R&D Antenna Engineer:

 

SATIMO

Multipath fading is due to the radio signal reflecting off of various surfaces on its way to the receiver. These reflections create multiple signal paths as seen by the receiver (multipath) and these variations of the same signal constructively and destructively interfere with each other when they reach the receiver. The destructive interference causes drops in data throughput and, in extreme cases, a dropped connection. Constructive interference can blast the receiver with too much power, leading to signal distortion in the receiver front end.

    The biggest environmental concern is the shift to indoor wireless usage. Indoor is rapidly becoming the primary usage environment for wireless devices of all types. The close proximity of walls, fixtures, furniture, etc in a contained space creates a much higher level of multipath -- even to the point where a line-of-sight path to the base station is the exception, rather than the rule.

 

In real propagation channels, the direct link between the mobile station and the base station coexists with the indirect links caused by the reflection and the diffusion of this ray on the buildings, and different obstacles surrounding the emitters and the receivers. Those different rays are recombined in a constructive or destructive way: this create signal fading at the location of the terminal. Usually, buildings and their related indoor environments are the most common environmental concerns when deploying a cellular network.

Madhusudhan Gurumurthy – Senior Applications Specialist

 

Spirent Communications

Moray Rumney

Lead technologist

 

Agilent Technologies

A wirelessly transmitted signal undergoes multiple reflections which traverse diverse paths before reaching the receiver. Multiple copies of the transmitted signal arrive at the receiver with different delays, gains and phases. The vectorial combination of these copies results in frequency-dependent constructive or destructive interference. This constructive or destructive behavior is known as multipath fading.

 

Multipath fading is a consequence of the fact that wireless devices are not connected with wires. For most of the 100 years of wireless history multipath fading has been considered as an undesirable property of the propagation channel, until the concept of spatial multiplexing was developed. Spatial multiplexing was enabled by the growth in signal processing power which is required to disentangle the multiple data streams launched into a multipath channel. The acoustic analogy is like being able to tune into different conversations in a crowded room. Multipath fading is a complex subject and gets a great deal of attention but of perhaps equal environmental concern to the performance of MIMO systems is multipath interference. Most simulations model interference as broadband, isotropic and Gaussian whereas in any real OFDM environment it is narrowband, directional and statistical.

 

Q: How does antenna diversity help address multipath fading?

 

Derek Skousen – Product Marketing Analyst:

 

 

 

MI Technologies

Nicolas GROSS – Applications Director , Mr. Alessandro SCANNAVINI - Field Application Engineer and Meryam ABOU EL ANOUAR - R&D Antenna Engineer:

 

SATIMO

Fading occurs when reflections create multiple versions of a signal which combine, constructively or destructively, at the single reception point. Antenna diversity splits that single reception point into separate reception points. Even though each antenna will receive a multipath signal, if they are independent enough, they will receive it in different ways. Now the receiver can filter and combine these different inputs to its advantage. If you had a separate antenna for every signal path, you would maximise antenna diversity and could, in theory, eliminate the fading effects of multipath. In practice, much of the gains from diversity can be realized with just a few antennas.

 

Antenna diversity can be implemented by using different mechanism: spatial diversity, frequency diversity, angle of arrival diversity, polarization diversity, multipath diversity, and pattern diversity.

The most common and simplest mechanism is the spatial diversity. It consists of having two antennas separated by physical distance. Due to the phase delays, multipath signals arriving at the antennas differ in fading. This translates in having an improvement in SNR from a diversity antenna system over a single antenna system. This improvement is usually termed “Diversity Gain”.

 

Madhusudhan Gurumurthy – Senior Applications Specialist

 

Spirent Communications

Moray Rumney

Lead technologist

 

Agilent Technologies

Antenna diversity alleviates multipath fading by taking advantage of the phenomena that causes signal fading to begin with. When multiple antennas are available, there are numerous ways to select or combine signals from the antennas.    This may be as simple as periodically selecting the strongest signal, or it may involve changing the gain and phase of the signals received on each antenna and summing them together.  The general idea is to improve the received signal and avoid outages that fading would cause on a single-antenna.  Anything that helps to make the signals "look different" from one another reduces the chance that multiple antennas will experience the same fading at the same time.  Statistically the best performance can be obtained if each antenna sees fading that is independent from the fading seen by other antennas.

 

Antenna diversity in a multipath environment is like having two bites at the cherry. For maximum benefit the antennas will be pointing in different directions with different polarizations and spacing, a bit like the ears on your head. Each will pick up a different version of the incident multipath signals and multipath interference such that advanced receiver algorithms can process the data to maximize the received SINR. Antenna diversity therefore improves the robustness of the receiver and is most useful in low SINR conditions. It is distinct from spatial multiplexing which increases data rates using multiple data streams, which in turn requires significantly higher SINR.

 

 

 

Q: What is fade margin?

 

Derek Skousen – Product Marketing Analyst:

 

 

 

MI Technologies

Nicolas GROSS – Applications Director , Mr. Alessandro SCANNAVINI - Field Application Engineer and Meryam ABOU EL ANOUAR - R&D Antenna Engineer:

 

SATIMO

Fade margin is the amount of additional system gain or receiver sensitivity that must be maintained to allow for multipath fading effects.

 

Fading margin gives an indication of how many dBs a received signal level can be decreased in order not to impact receiver performances.

 

Madhusudhan Gurumurthy – Senior Applications Specialist

 

Spirent Communications

Moray Rumney

Lead technologist

 

Agilent Technologies

Fade margin is power margin that ensures the receiver can withstand a fade without experiencing a signal outage.

 

Fade margin can be looked at either from the perspective of the signal or the receiver. From a signal perspective it is the amount by which the signal level can be reduced without degrading the radio link beyond a specified performance. From the perspective of a receiver the fade margin is the range over which the receiver can follow the fading signal.

 

Q: How does one calculate NLOS path performance?

 

Nicolas GROSS – Applications Director , Mr. Alessandro SCANNAVINI - Field Application Engineer and Meryam ABOU EL ANOUAR - R&D Antenna Engineer:

 

SATIMO

Madhusudhan Gurumurthy – Senior Applications Specialist

 

 

 

Spirent Communications

In NLOS condition the signal might have undergone scattering, diffraction, polarization changes, and reflection. All of them affect the received signal.

There are NLOS models which could be used to predict the RF signal strengths.

These models provide estimates of the path loss considering distance between TX and RX, antenna gain, antenna heights, used frequencies, and surrounding environment.

 

The Non Line-of-sight scenario is the typical case in a cellular radio environment.  Measurements are typically made on devices using a specific channel model that represents a standardized condition.  Such a channel model typically includes a multipath power delay profile with Rayleigh fading on each delayed signal.  The overall power is adjusted to a prescribed level to measure the device performance.

 

Moray Rumney

Lead technologist

 

Agilent Technologies

 

There are several approaches to calculating non line of sight path performance. Simple ray tracing is conceptually simple but does not scale well to complex environments. Geometry-based models such as SCME and WINNER, are based on ray tracing but simplify each path to at most 20 rays. The geometry of each path is described in terms of its angular spread, AoD and AoA. The Doppler spectrum and spatial characteristics are built from multiple simulation drops with random phases for each ray. Correlation-based models take an explicit approach to describing the channel by defining the Doppler spectrum for each path. If the geometry of each path is the same for the two approaches, then the results are comparable provided the number of drops is large enough.

 

 

 

Q: What is the impact of the EUT on the test antenna type and quantity?

 

Nicolas GROSS – Applications Director , Mr. Alessandro SCANNAVINI - Field Application Engineer and Meryam ABOU EL ANOUAR - R&D Antenna Engineer:

 

SATIMO

Moray Rumney

Lead technologist

 

 

 

Agilent Technologies

EUT size constraints largely dictate the antenna configuration design and the antenna type to be used for MIMO implementation. Specifically, the EUT chassis could be the most critical factor when designing a multiple antennas system at low frequency, since the ground plane becomes the main radiator at that specific frequency, and the spatial separation of the antennas in terms of wavelength becomes limited, which directly impacts the correlation, and hence the MIMO performance.

 

There are many factors but size is probably the most obvious. From a user device perspective the easiest case is the upcoming customer premises equipment (CPE) specification being developed by 3GPP which involves the installation of a separate external high performance antenna system from the mobile terminal located inside a building. Moving down in size, a large laptop gives the potential to design several spatially diverse antennas into the lid of the device although great care still needs to be taken with PC-generated noise e.g. from the display. Large smartphones are the next jump down in size and finally dongles present the biggest antenna challenge, particularly for the 700 MHz band.

 

 

Q: What antenna separation is sufficient to achieve an acceptably low mutual fading correlation for a mobile terminal in a similar environment?

 

Derek Skousen – Product Marketing Analyst:

 

 

 

MI Technologies

Nicolas GROSS – Applications Director , Mr. Alessandro SCANNAVINI - Field Application Engineer and Meryam ABOU EL ANOUAR - R&D Antenna Engineer:

 

SATIMO

The easy answer is often "a half wavelength", but the practicalities of design requirements make any easy pronouncements useless – especially since there are more factors involved than separation. For example, the effects of orientation and coupling of other nearby components (both dielectric and conductive) strongly affect correlation and efficiency performance.

 

When talking about “spatial diversity” a rule-of-thumb is to have the antenna elements spaced 0.5λ. This will guarantee both an antenna correlation lower than 0.5, and low mutual coupling. However, the required element spacing is strictly related to the design and the type of antenna being used.

 

Moray Rumney

Lead technologist

 

Agilent Technologies

 

Low correlation can be achieved either by physical separation or by polarization. The continued shrinking of device sizes means that for many mobile devices, particularly those operating below 1 GHz, it is impractical to separate the antennas sufficiently for low correlation. Adopting cross polarization allows for a more compact design which will work well in an environment with significant polarization scattering. That said, it was mentioned earlier that the practical correlation possible with realistic head and hand loading mean that it may be better to optimize for antenna efficiency and gain imbalance rather than low correlation.

 

 

Q: What is Adaptive MIMO Switching?

 

Moray Rumney

Lead technologist

 

Agilent Technologies

Madhusudhan Gurumurthy – Senior Applications Specialist

 

Spirent Communications

In LTE, MIMO is represented by six different downlink multi-antenna transmission methods ranging from transmit diversity through spatial multiplexing to beam-steering. Adding the default SISO makes for a total of seven transmission modes. In the uplink MIMO exists only in multi-user form since multiple UE transmitters are not defined until Release 10. Each transmission mode has particular benefits in different radio conditions. An optimized network will switch between different modes according to the prevailing or instantaneous conditions. The specifications take a toolbox approach to MIMO adaptation in that they provide the mechanisms for switching between modes (some being faster than others) but not the algorithms for determining the optimum choice. This leaves the door wide open to proprietary optimization by network vendors.

There are several different MIMO transmission schemes, each of which function with different spectral efficiencies and throughputs under different conditions. Some of the factors affecting performance of a MIMO system are channel conditions (correlation, amount of scattering, antenna orientation), SNR and vehicular speed, all of which change continuously. MIMO systems adapt to these changing conditions by selecting the transmission schemes that achieve the best throughput under each set of conditions. This is Adaptive MIMO switching (AMS). For example, LTE has 7 different transmission modes. The mode-switching algorithms are proprietary and their abilities to adapt to dynamic environment conditions will be a key differentiator between MIMO solutions.

 

 

 

Main
My Photo

Categories

Blog powered by TypePad
Member since 11/2008

Other Horizon House Sites:

Microwave Journal Online: Home | Current Issue | News | Buyer's Guide | Events | Resources | Archives | Subscriptions | Privacy Policy

Advertiser Information:
2009 Media Planner

Find out why more companies advertise in Microwave Journal than any other publication in the industry.

Read More >>

Microwave Journal
Editorial Information

Editorial Planning Guide and Information for Authors

Read More >>


©2009 Microwave Journal & Horizon House Publications ® All rights reserved.