Proceedings of IC-NIDC2014
Joint Optimization of Base Station Clustering and
Beamformer Design in Cloud-RAN Networks Under LTE
System
Dexian Ma
1
, Ping Gong
2
, Kai Niu
3
Key Laboratory of Universal Wireless Communication,Ministry of Education
Beijing University of Posts and Telecommunications, Beijing 100876, China
Jaunary_1990@126.com, pgong@bupt.edu.cn, niukai@bupt.edu.cn
Abstract: In this article, we consider a downlink Cloud-
Radio Access Network (C-RAN) with a central
processing unit, distributed Remote Radio Heads (RRH),
a high speed and low latency front-haul which connects
RRHs to the central processing unit and amounts of
mobile users (MS). By exploiting the centralized
processing of the C-RAN system and the advantage of
breaking the traditional cell restrictions, we propose a
method to jointly optimize the base station clustering and
beamformer together to maximize the system capacity in
LTE system. In this article, we formulate the problem
with a sparse question. Due to the large difficulty of
solving the sparse question, we address it by transforming
the target function into an equivalent solvable form. The
simulation results show that compared with the
Coordinated Multiple Points (CoMP) system, the
proposed method in C-RAN scenario has almost 40%
performance improvement. At the same time the
convergence of the method is guaranteed.
Keywords: C-RAN; Base station; JP; Cluster;
Beamformer; MMSE
1 Introduction
C-RAN is a radio access network (RAN)
implementation and deployment solution with great
promise to solve the problem in current RAN deployment.
With the surge of the mobile traffic, C-RAN becomes an
innovative architecture to provide better services to the
customs. C-RAN networks architecture is based on the
distributed base station type which consists of RRHs and
BBUs. BBUs are a baseband pool composed by several
real-time virtual baseband processing units and RRHs are
distributed deployment in the system, each one of which
doesn’t belong to any specific physical BBU [1] [2].
Unlike traditional distributed base station architecture, C-
RAN breaks up the static relationship between RRHs and
BBUs. In this work, we focus on effectively maximizing
the capacity of C-RAN and meanwhile minimizing the
system overhead and complexity.
To improve the capacity of the C-RAN system has
been a challenging topic of current researches. In
coordinated transmission and reception system, there are
two main methods: 1) joint processing (JP) 2) coordinated
beamforming (CB). Employing the JP transmission
method throughout the whole network can achieve the
best performance of the system because each one of the
base stations knows the information of others very well
[3]. At the same time, the inter interference of the base
stations can be canceled out by jointly precoding.
Unfortunately, the tremendous signaling overhead of the
system makes this approach impractical. When the
benefit of complete JP among the base stations is
outweighed by the signaling overhead, the base stations
can choose partial JP. That means applying full JP among
the base stations in one cluster as an alternative in order
to reduce overhead of the system [4]. Meanwhile jointly
optimizing the beamformers is applied among the
coordinated base stations to suppress interference. In the
C-RAN architecture, the characteristics of the centralized
and coordinated baseband processing among base stations
makes this scheme feasible [5].
As mentioned above, we adopt to use partial JP to
improve the system capacity. In previous studies, many
works attempt to optimize partial JP for different network
configurations. But most of them have a limitation, that
they handle the base stations clustering and beamformer
separately [6] [7]. Due to the sequence of the clustering
and beamformer, they can obtain only a suboptimal
solution rather than the global optimal solution.
In our work, we consider C-RAN scenario in
downlink LTE system. Taking advantage of the C-RAN
system characteristics. We adopt the approach that jointly
optimizes the clustering and beamformer. The RRHs
distributed in the C-RAN network will dynamically form
the collaborative clusters, which are composed by a small
amount of the RRHs. In addition, the RRHs will also
apply JP transmission mode in each one of the clusters,
which guarantees the overhead of the system under the
limit of the front-haul capacity in conjunction with the
beamformer design to achieve the best system
performance. For the purpose of further reduction the
system overhead, we hope that the precode matrix
contains zero vectors as much as possible. That structure
is called the group-sparse structure. In order to jointly
optimize the beamformer, we penalize the group-sparse
structure with a mixed
21
/ll
norm. For the reason that the
optimal function is hard to solve, we transform the
optimal function into a solvable form which makes the
problem solvable.
资源评论
