This Thesis starts in the framework of a research project supported by the Ministerio de
Educación y Ciencia. This research project analyzed a new radio access architecture for UMTS systems,
the Multinodo B architecture. The key idea for this new proposal for UMTS systems was to spatially
distribute the antennas from the base station around the service area, with the aim of reducing the
interference of the system and thus, increasing its capacity.
On the other hand, communication systems with multiple antennas or MIMO systems
(Multiple-Input Multiple-Output), have received increasing attention from the research community
over the last ten years. Initially proposed for point-to-point links, MIMO systems are seen as an
interesting option for future communication systems due to their high spectral efficiency.
In this Thesis, the proposed Multinodo B architecture is analyzed and characterized from
the point of view of a coordinated multi-cell multi-user MIMO system. In this advanced and
cooperative MIMO configuration, base stations achieve perfect data cooperation through a high
speed backbone and hence, there is a chance for transmitting coordinatedly to all users in the area.
Moreover, per-base station power constraints are taken into account.
The analysis of coordinated multi-cell multi-user MIMO systems is still in its first stages.
Nowadays, it is possible to differentiate between two research lines; on one hand, some progress is
taking place in order to obtain the capacity region of this MIMO configuration and on the other
hand, several algorithms are being proposed to solve different optimization problems.
In the Thesis, the power minimization problem is stated for the downlink of a coordinated
multi-cell multi-user MIMO system. The power minimization problem is an optimization problem that
tries to minimize the total transmitted power in the system, which can be defined as the sum of
the transmitted power from all the base stations in the system. At the same time, all the users
are subject to a certain quality of service requirement.
Firstly, a system model is proposed, where the process carried out at each receiver is
omitted. In this context, a first Joint Power Control and Optimal Beamforming (JPCOB) algorithm
is stated by modifying one of the classical references dealing with the power minimization problem.
This first algorithm, the JPCOB-VUL or Virtual uplink-based JPCOB, makes use of the duality between
the power minimization problem in the downlink and an equivalent virtual uplink formulation for
simplifying the transmit beamformer design. Moreover, the structure of the matrices involved
in the power control mechanism allows to formulate an efficient version of the JPCOB-VUL algorithm.
In the analysis of some practical concerns dealing with the JPCOB-VUL algorithm, a model
for the interferences present in the system is proposed. Moreover, several spreading sequence
assignment schemes are presented in order to minimize the interference power received by the users
in the system. On the other hand, a robust version of the JPCOB-VUL is obtained for the cases in
which base stations have partial channel state information in the form of hybrid channel knowledge.
This model of channel information stands for the case in which there is a constraint in the amount
of information that base stations can exchange through the high speed backbone.
Secondly, a complete and matricial system model is proposed, which totally characterizes
the main features of a coordinated multi-cell multi-user MIMO system, including realistic system
parameters such as the asynchronous nature of the signals arriving at each user and the process
carried out by the receivers. At this point, a JPCOB algorithm is proposed which outperforms the
JPCOB-VUL algorithm, the JPCOB-DL or Downlink-based JPCOB. Moreover, a simplified version of this
proposal, the JPCOB-SDL or Simplified Downlink-based JPCOB, is obtained for limited feedback
scenarios, where there is a constraint in the amount of information that each user can feedback
to the base stations. JPCOB-DL and JPCOB-SDL algorithms are regarded as downlink-based algorithms
because both algorithms do not make use of the duality between the downlink and virtual uplink
problem.
Finally, two schemes are proposed in order to improve the convergence properties of the
JPCOB-DL algorithm in ill-conditioned coordinated multi-cell multi-user MIMO systems. Simulation
results show that the proposed schemes can be seen as efficient and computationally simple
criteria for removing users from a coordinated system in case the system is not feasible.