RF Beamforming Techniques Improve TD-LTE Cell-Edge Performance
By Craig Grimley, Agilent Technologies
A common theme in modern wireless communications is the use of multi-antenna techniques, such as beamforming and spatial multiplexing in order to improve the cell capacity and throughput. This article will briefly summarize multi-antenna techniques, before introducing the concept of beamforming, along with its advantages and specific use within the roll-out of 3GPP LTE (3rd Generation Partnership Project Long Term Evolution), and particularly its TDD (Time Division Duplex) variant. A discussion of multi-antenna beamforming measurement challenges from an eNB base station test perspective will also be provided, which will highlight the importance of calibration when it comes to testing the performance of a beamforming transmission system.
To recap, the basic radio channel uses a single transmit and single receive antenna and is known as SISO (Single Input, Single Output). This simple radio channel sets the baseline for transmission performance against which all more complex transmission configurations can be measured.
SIMO (Single Input, Multiple Output) provides receive antenna redundancy compared to the SISO baseline, allowing the use of receive diversity techniques such as maximum ratio combining in the receiver. This improves SINR observed at the device receiver, and can help improve performance under channel fading conditions.
MISO (Multiple Input, Single Output) provides transmit antenna redundancy, allowing the use of transmit diversity techniques such as Alamouti symbol coding, or Space Frequency Block Coding (SFBC) as is the case for LTE. Similar to SIMO this also provides an improvement in the observed SINR at the device receiver, and can again help provide protection against channel fading.
Neither SIMO nor MISO offer data throughput improvement except in so far as they reduce error ratio and hence reduce the need for data retransmission.
MIMO (Multiple Input, Multiple Output) provides both additional transmit and receive antenna redundancy. If the same data is sent to the transmit antennas, this redundancy is used to improve the SINR at the device receiver using the same transmit and receive diversity techniques described above. Alternatively some or all of the potential SINR performance improvement can instead be traded off in order to obtain an improved spectral efficiency.
Spatial multiplexing transmission techniques, where the transmit antennas are used to send independent data streams, can either deliver increased data throughput for a single user (SU-MIMO or Single-User MIMO), or increased system cell capacity (MU-MIMO or Multi-User MIMO).
In addition to these diversity and spatial multiplexing techniques, it is possible to use multi-antenna configurations to focus transmission or reception in a particular direction. This can be used to improve system performance, can be either fixed or variable depending on the application, and is known as beamforming. Beamforming techniques are used in many different applications and frequencies from sonar and seismology, through acoustics to wireless communications, radio astronomy and radar.
In general, transmit beamforming works by exploiting the interference patterns that exist whenever the same signal is transmitted from two or more spatially separated transmission points. A similar principle applies whenever the same signal is received from two or more spatially separated reception points, which is exploited by receive beamforming techniques.
As a simple example, let’s consider the case of an RF wireless signal transmitted from a single omnidirectional antenna, the resulting signal relative field strength is shown in Figure 1A represented as a solid blue line.
To enable transmit beamforming a second identical omnidirectional antenna element is added, separated from the first element by half the RF carrier wavelength as shown in Figure 1B. In this example both antenna elements carry identical copies of the signal information symbol to be transmitted. Immediately it can be seen that in azimuth directions around 0 degree azimuth where constructive (or in-phase) interference occurs, the combined field strength increases, producing an effective coherent signal power gain in this direction. In contrast, azimuth directions around +/-90 degrees, where destructive (or out-of-phase) interference occurs, show a decreased or attenuated combined field strength.
Adding a third antenna element, separated along the same axis as the first two elements by half the RF carrier wavelength improves the spatial selectivity of the combined relative field strength, as shown in Figure 1C. In our example, the array elements are co-polarized, correlated and uniformly separated along a single antenna element axis, creating a Uniform Linear Array (ULA) antenna system. The formation of a single main lobe in the azimuth direction of 0 degrees relative the ULA broadside can clearly be seen. This is where maximum constructive (or in phase) interference occurs, producing a power gain maximum within the combined field strength beam pattern. The formation of two distinct power attenuation nulls, one on either side of the main lobe located at +/-42 degrees azimuth can now be observed. These two power minimum locations represent the azimuth directions in which maximum destructive (or out-of-phase) interference occurs within the combined field strength beam pattern.
Finally, adding a fourth antenna element to the ULA further improves the main lobe selectivity as shown in Figure 1D. The number of power nulls has also increased from two to three. Two nulls are now located at +/-30 degrees azimuth, with the third located on the ULA antenna axis line. The formation of two distinct power side lobes is now clearly observed, located at +/-50 degrees azimuth. Both side lobes appear at reduced power levels relative to the main lobe.
The resultant beam pattern is not only determined by the ULA physical geometry and element separation, but is also affected by the relative magnitude and phase weightings applied to each information symbol copy transmitted on each antenna element.
This can be demonstrated by now introducing a +90 degree relative phase shift weighting across each of the four antenna elements. The result is a shift of the main beam location from 0 degrees azimuth to -30 degrees azimuth as shown in Figure 1E. Note that the null and side lobe locations have also been affected by the new weighting values.
By careful design of the beamforming antenna array geometry, plus accurately controlling the relative magnitude and phase weightings applied to each of the antenna elements, it is possible to control not only the selectivity shape and azimuth direction of main lobe power transmissions, but also to control the power null azimuth locations and side lobe levels.
Let’s now separately consider the impact of adding additional antenna elements on the effective power gain of the resultant beam pattern observed at a target device receiver.
Figure 1B showed the addition of a second antenna element which transmitted an exact symbol copy of what was being transmitted on the first antenna element. In this case, the constructive in-phase signal summation would result in a 6 dB coherent power gain improvement observed by a target device receiver positioned at the 0 degree azimuth main beam location. Therefore if plot normalization were not applied, the main lobe maximum of Figure 1 plot B two antenna case would in theory be twice the main lobe maximum of the plot A single antenna case.
This 6 dB coherent gain improvement can be considered as the beamforming gain improvement observed at the target device receiver, due to using two spatially separated antenna elements relative to a single antenna transmission.
In practice, the symbol power levels transmitted on each of the two antenna elements may be reduced by 3 dB to half the original single antenna symbol power level, maintaining the same total transmitter power as the single antenna case. Even so, this would still result in a 3 dB beamforming gain observed at the target device receiver relative to a single antenna transmission.
The use of multi-antenna beamforming transmission is very attractive in modern wireless communication systems due to the combined advantages of beamforming selectivity, interference management and coherent signal gain.
Let’s summarize some important aspects and terminology used to describe beamforming transmissions in the context of Figure 2:
• Main Lobe: the primary maximum transmission power lobe, usually directed at the target device or a transmission path that will reach the target device by reflections in the radio propagation channel.
• Side Lobe: the secondary power transmission lobes which can potentially produce unwanted interference to other user devices within the serving or adjacent cells.
• Power Null: locations of minimum power within the transmission beam pattern which the system may choose to exploit and control in order to mitigate interference to other devices within the serving or adjacent cells.
• Main Beam Width (Φ): selectivity of the main lobe transmission measured as the degree azimuth spread across the 3 dB points of the main lobe.
• Main Lobe to Side Lobe Levels: the selectivity power difference of the desired main lobe transmission power relative to the unwanted side lobe transmission power.
One of the biggest challenges in any modern wireless cellular communications system is the performance at the cell edge. This is a major reason why beamforming technology has a key role to play in delivering LTE services. Figure 3 illustrates two practical scenario examples which both exploit the advantages of beamforming to improve performance within a modern cellular wireless communication system.
Figure 3A depicts two adjacent cells each communicating with a respective UE located at the boundary between the two cells. The illustration shows eNB1 is communicating with target device UE1, with the eNB1 transmission using beamforming to maximize the signal power in the azimuth direction of UE1. At the same time it can be observed that eNB1 is attempting to minimize interference to UE2 by steering the power null location in the direction of UE2.
Similarly eNB2 is using beamforming to maximize reception of its own transmission in the direction of UE2, whilst minimizing interference to UE1. In this scenario, it is clear that the use of beamforming can provide considerable performance improvements particularly for cell edge users. The beamforming gain can also be used to increase the cell coverage where required.
Figure 3B depicts a single cell (eNB3) communicating simultaneously with two spatially separated devices (UE3 and UE4). Since different beamforming weightings can be applied independently to each of the spatial multiplexing transmission layers, it is possible to use Space Division Multiple Access (SDMA) in combination with MU-MIMO transmissions in order to deliver an improved cell capacity.
Two different beamforming implementation techniques are illustrated in Figure 4. Figure 4A shows an example of a fixed conventional switched beamformer consisting of an eight-port Butler matrix beamforming network. This network implementation consists of a matrix of different selectable fixed time or phase delay paths, implemented using a combination of 90° hybrid couplers and phase shifters.
The number of fixed transmission beams produced is equal to the number of antenna elements N used to form the Butler matrix network. (The example shown uses eight antennas, producing eight selectable beams). This is sometimes also referred to as a “grid of beams” beamforming network, and supports selection of any individual or combination of the N fixed transmission beams in order to maximize the SINR at the device receiver.
In a wireless network, optimal eNB downlink transmission beam selection would primarily be driven by some knowledge of the UE position within the cell. This knowledge can readily be obtained directly through measurement of the uplink signal Angle of Arrival (AoA) across the eNB receive antenna array, or indirectly derived from uplink control channel quality feedback information.
In contrast, Figure 4B shows an example of an adaptive beamformer. As the name suggests, an adaptive beamformer has the ability to continually adapt and re-calculate the optimum applied transmission beamforming complex weighting values in order to best match the channel conditions.
Because the adaptive beamformer weightings are not fixed, it can not only optimize the received SINR at the target UE, but also better adapt the selectivity and power null positioning to minimize interference to other users.
In a wireless network, the eNB would typically estimate the optimum weightings through direct measurement of the received uplink reference signals observed across the eNB receiver array. This information can then be used to calculate the uplink Angle of Arrival (AoA) as well as decompose the channel characteristic matrix.
In the case of a Frequency Division Duplex (FDD) system, where the downlink and uplink use different RF carrier frequencies, the applied beamforming transmission complex weightings will be primarily driven by measured AoA information derived for both the target UE, as well as knowledge of any other UEs within the cell. Weighting estimation may also be aided by channel feedback information reported by the UE on the uplink.
For the case of a Time Division Duplex (TDD) system, since downlink and uplink share the same RF carrier frequency, channel reciprocity may be assumed. For this reason beamforming in a TDD system can outperform what is possible in an FDD system. The applied beamforming transmission complex weightings can be chosen to best match the decomposed channel characteristic matrix eigenvectors, as derived from the eNB received signal. These channel-matched beamforming weightings can help optimize the SINR observed at the target UE receiver. The eNB is not reliant on channel feedback information supplied by the user device on the uplink, although in practice, channel feedback may still be used in the eNB beamforming weighting estimation process.
Beamforming in LTE
LTE defines several downlink transmission modes to support beamforming. Of particular interest are Transmission Modes 7, 8 and 9. 3GPP Release 8 introduced TM7, which supports single layer beamforming. Release 9 added TM8, which supports dual layer beamforming and Release 10 added TM9, which supports up to eight layer transmissions.
Figure 5 shows a typical eNB RF antenna configuration used in TD-LTE cellular networks that support TM7, TM8, and TM9 MIMO beamforming modes.
The example is an eight-element physical antenna, configured with two groups of antenna elements. Each group is orthogonally cross-polarized at 90° to the other group. Antenna group 0 consists of antenna elements 1 through 4, polarized at plus +45°. Antenna group 1 consists of antenna elements 5 through 8, polarized at -45°.
Each of the elements within a given group are spatially separated by approximately half the RF carrier wavelength. This provides a high degree of antenna element correlation within the antenna group, which is good for coherent beamforming. Since each of the two groups are cross-polarized relative to each other, there is a low correlation between each of the two antenna groups, which is good for spatial multiplexing. Thus a typical TD-LTE eNB RF antenna physical configuration attempts to satisfy the desirable but conflicting correlation requirements for MIMO spatial multiplexing and coherent beamforming.
Typical TD-LTE eNB Beamforming Test System Configuration
One of the main test challenges for beamforming is the need to verify and visualize the beamforming signal performance at the physical RF antenna array, in order to validate the following:
• eNB RF antenna calibration accuracy
• Baseband encoded beamforming weighting algorithm correctness
• MIMO single and dual layer EVM at the RF antenna
The test system shown in Figure 6 uses the Agilent N7109A Multi-Channel Signal Analyzer and 89600 VSA software installed with TD-LTE measurements. The multi-channel signal analyzer can support eight phase-coherent RF measurement channels and, along with the appropriate RF splitters and attenuators, can easily be integrated into a typical TD-LTE base station test setup.
The key requirement for making good measurements is system calibration. A correction wizard guides the system calibration process, prompting the user to connect the signal analyzer channel 1 measurement cable to the first output port of the two-way calibration splitter at the injection point represented by a dotted line in Figure 6. All the cross-channel characterization measurements will be made referenced to channel 1. The user is then prompted to connect each of the remaining channels 2 through 8 measurement cables (located on dotted line) one at a time to the second output port of the two-way calibration splitter. In this way the correction wizard is able to characterize the cross-channel corrections required to compensate the signal analyzer beamforming measurements for all mismatch effects inherent in the measurement cables, connectors, splitters, and attenuators. As a result direct, corrected measurements of the antenna beamforming performance can be observed at the RF antenna output. The importance of test system calibration of magnitude and phase variations due to RF cabling and connectors cannot be overstated.
The VSA software and multi-channel signal analyzer are first used to display the time-synchronized RF signal capture from all eight antenna elements as shown in Figure 7. Any fundamental RF power or timing performance impairments can be identified quickly, before the more advanced demodulation measurements are attempted.
The VSA software TD-LTE measurement application provides a rich set of demodulation results for verifying downlink MIMO beamforming signals. These include IQ constellations, EVM result metrics, detected resource allocations, UE-specific RS weights, cell-specific RS weights and impairments, and UE-specific and common broadcast antenna beam patterns. Examples of some of these results are shown in Figures 8 to 11.
The demodulated IQ Constellations are displayed per spatial multiplexing layer, as shown in Figure 8 traces A and L, and provide a quick visual indication of the signals modulation quality correctness.
The frame summary shown in Figure 8 trace D provides access to individual EVM and power metrics associated with each channel and signal type. It also provides a color key for all channel type results, which is reused throughout the VSA traces.
The detected allocations displayed in Figure 8 trace B shows the resource block allocations for each user-specific transmission, plus resource allocations used by common control channels.
Measured UE-specific RS weights are presented in table format for each of the eight antenna elements, as shown in Figure 9. Weightings can be evaluated in both magnitude and phase down to the individual resource block allocations associated with each user transmission.
Separate UE-specific RS weights traces are available for each spatial multiplexing layer.
Measured cell RS weightings produce the blue cell RS beam patterns shown in the Figure 11.
Summary and Conclusions
The performance issues associated with any modern wireless cellular communications system are highest at the cell edge. This is the region where user devices experience the most degraded signal to noise conditions as well as highest levels of inter-cell interference. The use of multi-antenna beamforming transmission technology has a key role to play, particularly in TD-LTE networks where the uplink and downlink frequency is the same and channel reciprocity can be assumed. The combined benefits of beamforming selectively, interference management and coherent signal gain can help ensure a more consistent end-user experience, with delivery of key services at an acceptable level of performance across the entire cell.
From an eNB development perspective, the use of multi-antenna beamforming transmission presents some specific test challenges. These include the need to verify correct implementation of eNB baseband receive/transmit algorithms used to generate beamforming weightings, as well as accurate validation of eNB calibration performance observed at the RF Antenna. When testing a beamforming transmission system, care must be taken to correct for the physical measurement configuration setup used. In addition, since beamforming is combined with spatial multiplexing techniques there is a need to verify EVM performance of each MIMO layer observed at the RF Antenna.
If you want to see a demonstration of the measurements described above, there’s a demonstration video at www.youtube.com/watch?v=mj58aSOZ1Kc
this article to a friend!