Uppsala universitet

5G Wireless Research


This research at the Signals and Systems group has as its general aim to go beyond conventional cellular transmission, to enable an increased spectral efficiency. We focus on schemes that are useful also for mobile and moving terminals.

As a result, we have so far suggested new solutions and provided new concepts and tools in three areas:

1. Coordinated multipoint transmission and cell-free (massive) MIMO downlinks

Coordinated Multi-Point in Cellular Networks: From theoretical gains to realistic solutions and their potentials (slides in pdf, from tutorial at e.g. IEEE ICC.)

Shadowed areas and interference at cell borders pose challenges for future wireless broadband systems. A potentially powerful remedy is joint transmission in downlinks from several cells. Coordination can avoid interference and joint non-coherent transmission can improve the received power. We may abandon the cellular concept and instead combine transmission resources that are most appropriate for each user. The most powerful scheme, joint coherent transmission, also denoted Network MIMO also cancel interference by phase cancellation. We have focused on downlinks (joint transmission), which generate more difficult challenges than uplinks (joint reception).

A main outcome here is that network MIMO, which has encountered considerable skepticism, is actually a realistic proposition. It can be made to provide large performance gains, but you have to construct the solution carefully.

In particular, cooperation areas (that could represent clusters of cells that perform joint downlink transmission) have to be designed carefully, to provide gains to most users. Groups of users that cooperate in a time-frequency resource block need to be selected well, but still fast and efficiently. Good channel prediction is essential since the whole scheme is rather sensitive to channel estimation errors and channel outdating.

Our proposed design for downlinks in homogenous networks combines Kalman channel prediction, robust linear joint precoding and appropriate grouping of users who are to cooperate in a particular time-frequency block. This will provide significant gains for most stationary and pedestrian users that are interference-limited. For high-speed users, channel prediction errors become prohibitively large. For these users, we propose to use non-coherent joint transmission.

The tools we have developed for the design of precoders (joint beamformers) range from near-optimal global optimization by Particle Swarm optimization to very low-complexity iterative robust MMSE designs. Very low-complexity user grouping schemes have been shown to be very efficient. We have also analyzed the influence of the properties of the control channels and the backhaul network, and included backhaul restrictions into the precoder designs. Such aspects, as well as the reliability of channel state information and the utility on the application level of being assigned a better channel, are relevant when selecting what users should be served by CoMP and what users are better served by cellular transmission.

We have also studied the use of CoMP also in heterogeneous networks and the joint use of massive MIMO, CoMP and small cells. These tools can be used to improve coverage, improve the throughput and also to the outage statistics and thereby reduce the latency of transmission.

For more information on our results on joint processing, please see our list of publications.

2. Low-overhead channel estimation for FDD massive MIMO (and multipoint) downlinks

Overview Slides

In downlinks of wireless broadband systems (5G and beyond), we today need channel state information (CSI) at the transmitter for link adaptation and scheduling, and in the future more precise information will be needed for downlink adaptive beamforming and coordinated transmission.

In frequency division duplex (FDD) downlinks, this requires many complex-valued channel gains to be estimated based on known downlink reference signals, with acceptable training (and feedback) overhead. This is one of the main remaining open problems in the research on massive Multiple Input Multiple Output ( MIMO) systems.

We have here presented a solution, based on combining several concepts. First, massive MIMO downlinks use a fixed grid of beams. For each user, only a subset of beams will then be relevant, and require estimation. Second, sets of coded reference signal sequences, with cyclic patterns over time, are used. Third, each terminal estimates its most relevant channels.

For each user equipment (UE), the most important of hundreds of channels can then be estimated with a low referernece signal overhead of 4% -10%. As a result, Massive MIMO and multi-cell cooperation is enabled for FDD.

Paper, with more details

3. Moving networks and improved links to vehicular users

Communication to users in vehicles is becoming increasingly important. Trains, buses and trams provide natural hot-spots with users who would like to use data-intensive applications, in particular streaming. The same is increasingly true for passenger cars. We then have sets of users who move together at high velocity through a cellular system.

This situation is challenging, but also contains many potentials for improvements. Could we gain by performing handover to a new cell for all users in the group, instead of the users individually? What could be gained by placing an antenna system on the vehicle roof, and local antennas indoors, instead of forcing all users communicate with the base station directly?

In this area, we have seen that the outdoor-to-indoor penetration loss via (perhaps metalized) vehicle windows play a large role in determining the potential gain of alternative schemes, such as vehicular repeater systems or relay nodes.

Predictor Antennas:

Overview Slides

Our present focus is to radically improve the performance of antenna systems outside (on the roof of) vehicles. We would like to be able to use the most advanced adaptive transmission schemes here, like advanced link adaptation, adaptive beamforming, massive MIMO and CoMP. A challenge is that such schemes require channel state information at the transmitter Such information will be outdated due to transmission time delays at vehicular velocities.

We have here proposed and are analyzing a simple but very powerful idea: The "predictor antenna". Place an antenna that senses the radio environments in front of the main antennas on the vehicle roof. This has in measurements shown to increase the attainable prediction horizons for vehicular users by at least an order of magnitude, and promises to enable a significant increase in spectral efficiency in broadband transmissions to vehicular users.

For more information on our results on moving networks and predictor antennas, please see our list of publications.