Multi-antennas for improved LTE performance
The previously often neglected third dimension – elevation – is in focus for the latest research on multi-antennas. We’ve put together an overview of key multi-antenna components and techniques for you, and we’ll explain how they can be used to maximize performance of LTE networks when taking advantage of vertical as well as horizontal spatial dimensions.
LTE was designed to support multiple transmit and receive antennas from its very first release, something that profoundly increased the commercial interest in such techniques. In fact, with LTE the use of at least two antennas at the base station and two receiver antennas in the user equipment (UE) has become the norm.
Multi-antenna research and development at Ericsson dates back to the early nineties and includes pioneering demonstrations of multiple-antenna technologies in GSM, HSPA and LTE. In Ericsson Research, we continue to spend great efforts in covering the vast field of multi-antennas. Before we dive into the focus of recent research – the vertical aspects of our three-dimensional world – let’s take a look at some important and commonly used multi-antenna components. In LTE, these include spatial multiplexing, rank adaptation and UE specific beamforming:
By employing spatial multiplexing several data streams can be transmitted in parallel over the resulting multiple-input-multiple-output (MIMO) channel (Fig 1).
Figure 1: SU-MIMO where an adaptive number of data streams (rank adaptation) are spatially multiplexed over the same MIMO channel.
In the case of single-user MIMO (SU-MIMO), the multiplexed data streams belong to the same UE and so-called rank adaptation offers the possibility to dynamically adapt the number of data streams for a UE to current channel conditions. Another option – so far less common – is spatial multiplexing of the data streams of more than one UE. This is referred to as multi-user MIMO (MU MIMO). SU/MU-MIMO can be combined with UE specific beamforming, which directs the transmitted energy towards the user of interest, thereby increasing the signal level relative noise and interference and thus the performance. UE specific beamforming is a form of precoding and works by adjusting the phase of each antenna to achieve constructive superposition of received signals (Figure 2).
Figure 2: Precoding used for UE specific beamforming.
The basic MIMO techniques in LTE are often used on top of more classical antenna techniques. One example is sectorization, which divides the spatial domain into separate angular intervals associated to different cells. Sectorization can alternatively be seen as a form of MU-MIMO, but where each sector represents a fixed cell specific beam and where the dynamic scheduling in each sector is performed separately.
Coordinated Multi-point (CoMP) is another area we study in conjunction with multi-antenna techniques. Historically, different cells have for the most part been operated in a rather independent manner. This may lead to high interference levels on cell-edges when the traffic demands increase. CoMP –coordination of transmissions/receptions across cells – offers a concept for addressing this. For example, beamforming can be coordinated across cells so that aggressor cells purposely place nulls towards victim UEs of neighboring cells (Figure 3).
Figure 3: UE specific coordinated beamforming.
Coordination can be fast and dynamic; or slow; or even (semi-)static. Dynamic coordination on a millisecond basis has the ability to closely track the time-varying traffic and may be an integral part of the scheduling, including performing coordinated beamforming in a UE specific manner. At the other end we have slower coordination techniques, for example cell shaping. Here, beam shapes and pointing directions are adapted in a coordinated manner across cells (Figure 4) for cell defining signals, such as synchronization and cell specific reference signals. This implicitly alters which cell serves the UE and thereby provides a tool for reducing interference as well as balancing the load among cells.
Figure 4: Cell shaping changing the coverage areas to avoid a cell edge in hotspot of UEs.
So what about the 3D aspects?
Traditionally and still, evaluations in the wireless communication field use channel models with only two dimensions, even though we live in a three-dimensional world. The vertical direction is basically non-existent in these models, all UEs are assumed to be placed on ground-level. This is finally about to change in an official sense as 3GPP has recently defined a new 3D channel model: 3GPP TR 36.873, “Study on 3D channel model for LTE” that also takes elevation aspects into account.
UE specific elevation beamforming is one key technique that we are exploring in the context of 3D channel models. It allows a beam to be directed in a way that suits each individual UE in the cell. For example, a UE high up in a high-rise may desire a beam pointing upwards, while a UE on the ground level may get a downwards pointing beam (Figure 5). This is in contrast to conventional systems with fixed tilt angle where a single beam serves all UEs in the cell.
Figure 5: UE specific elevation beamforming.
UE specific elevation beamforming complements the traditional way of performing beamforming purely in the horizontal direction, which however continues to be an interesting and powerful technique on its own. UE specific elevation and horizontal beamforming may also be combined, leading to two-dimensional (2D) beamforming. Also the other previously mentioned multi-antenna techniques can exploit the elevation domain. For example, the elevation domain can be utilized for vertical sectorization (Figure 6) or combined with higher order sectorization in the horizontal domain to perform 2D sectorization.
Figure 6: Sectorization in the vertical domain.
The related MU-MIMO technique can be used to co-schedule UEs that appear in different horizontal and/or elevation angles. Coordinated beamforming may utilize the additional degrees of freedom provided by the elevation domain to more efficiently avoid interference on victim UEs. There are almost endless opportunities in combining the various basic multi-antenna components.
The performance potential of beamforming techniques tends to increase with an increasing number of antennas, since the baseband gets access to more degrees of spatial freedom. This is facilitated by techniques for active antenna systems (AAS) where the radio is integrated into the antenna so as to offer possibilities for finer grained digital control of the beamforming weight of each individual subelement within the antenna.
Consider, for example, what is conventionally referred to as a “4 Tx/Rx” antenna array. This consists of two columns, each with 8 rows of cross-polarized subelement pairs (Figure 7). Only four ports would then be available for the spatial processing. But with the most extreme form of AAS, there would be one radio transceiver for each subelement meaning that the digital processing would have access to 4*8 = 32 ports of spatial degrees of freedom!
Figure 7: A classical antenna with 4 analog ports offering 32 digital ports with most extreme AAS.
Such an extreme form of AAS may not always be the most practical but it provides an illustration of how the spatial degrees of freedom may be multiplied, while more or less keeping a similar overall size of the antenna installation.
The possibility to use AAS is not the only reason why antenna installation size may constitute less of a hurdle for the exploitation of multi-antenna techniques in the future. The expected use of progressively higher frequencies reduces the size of each antenna element thereby facilitating maintaining reasonable installation size. Higher frequencies also tend to have reduced coverage and techniques such as beamforming may then help in maintaining coverage.
The future of multi-antennas techniques thus looks bright and Ericsson is proud to be at the forefront in the work on researching and developing this area. We have been a strong driver for the successful development and adoption of multi-antenna solutions in widespread wireless cellular standards such as LTE and HSPA and you can bet on that we will stay in the lead in the future as well! :)
George Jöngren Ericsson Research