Joint quantisation levels and power
optimisation in uplink network multiple-input
and multiple-output under constraint
backhaul
ISSN 1751-8628
Received on 19th January 2015
Revised on 17th June 2015
Accepted on 17th July 2015
doi: 10.1049/iet-com.2015.0358
www.ietdl.org
Shipeng Wang
✉
, Li Chen, Xiaohui Chen, Guo Wei
Key Laboratory of Wireless-Optical Communications, Chinese Academy of Sciences, School of Information Science and Technology,
University of Science and Technology of China, No. 96 Jinzhai Road, Hefei, Anhui Province, 230026, People’s Republic of China
✉ E-mail: shpwang@mail.ustc.edu.cn
Abstract: This article studies the performance of the cooperative system where each base station (BS) is connected to a
central process or (CP) via a noiseless and rate-limited backhaul link. The single-user compress-and-forward scheme is
deployed at the BSs, and the problem for jointly setting the quantisation noise and the transmit power levels is left
open. The authors formulate the weighted sum-rate (WSR) problem of optimi sing quantisation and transmit power
levels for K-users and B-BSs situation. However, the problem is non- convex and difficult to be sol ved directly by
conventional convex methods. In this work, an algorithm is proposed to solve the optimisation problem and achieve its
global optimality. More importantly, the authors reveal the reason for the effect of the quantisation levels on the
performance of cooperative system. Lower and upper bounds of quantisation levels are derived u nder the given signal
to interference noise ratio (SINR) requirements. The lower and upper bounds are then demonstrated to be rather close
to the numerical results. It is also numerically shown that the proposed algorithm can achieve the global optimality
and is robust with respect to different multi-user situations.
1 Introduction
In modern cellular systems, network capacity is mainly limited by
inter-cell interference. The limitation is particularly severe near the
cell edge where the quality of service deteriorates sharply due to
strong interference from the nearby cells. To address this issue,
joint multicell processing (MCP) or cooperation of multi-point
(CoMP) is adopted as a superior approach to eliminate inter-cell
interference [1]. In the cooperation among BSs, joint decoding in
the uplink and joint transmission in the downlink are employed to
overcome the interference effect. Theoretically, the interference
can be eliminated entirely when BSs share all the data and perfect
channel state information (CSI) with each other. In the uplink
scenarios, MCP and CoMP are also called as network
multiple-input and multiple-output (MIMO).
The inchoate concept of network MIMO is mentioned by
the authors in [2]. The capacity region of network MIMO can be
easily obtained with infinite backhaul link capacity, because the
downlink scenario can be degraded to a broadcast channel (BC) and
the uplink network MIMO can be degraded to a multiple access
channel (MAC). However, in practical uplink scenarios, the effect of
finite backhaul capacity must be taken into consideration. Moreover,
owing to the finite backhaul links, the BSs need to compress
their received signals and then share the quantised signals with each
other via the finite-capacity backhaul links. Thus the quantisation
levels motivate us to explore their effect to the achievable rate
region of uplink network MIMO model with limited backhaul.
Researchers have studied uplink network MIMO with
finite-capacity backhaul in some of the latest research proposals. In
these researches, there are two major kinds of situations. One is
that the BSs are connected with each other for cooperation. While,
the other one is that the BSs are connected to a central processor
(CP) for cooperation. In the former situation, distributed
interference subtraction scheme is applied to acquire an achievable
data rate in [3]. Then in [4], a comprehensive analysis of rate
region under different cooperative schemes and different
quantisation scheme provides new insight into the value of
network MIMO. However, about the second situation, an
achievable rate from user to the CP is given based on a single-user
scenario in [5], and whether the BSs are capable of decoding or
not is also considered. In [6], the situation that two users
communicate with a CP via two BSs is considered. The author
also derives the upper and lower bounds of the average achievable
throughput. Further, under the Wyner model, the achievable
throughput of the multi-user situations are taken into consideration
in [1, 7]. While in [8], the performance of various multiuser
detection scheme is compared under a Gaussian interference
channel. Nevertheless, these works do not study the effect of the
quantisation levels to the performance of the system.
Some authors have also looked at the effect of quantisation levels
in [9–13]. In [9], the authors analysed the performance of the
cooperative system where complex distributed Wyner–Ziv (D-WZ)
compression was used. Further, a robust and efficient distributed
compression for cloud radio access network is proposed to cope
with uncertainties on the correlation of the received signals in
[12]. In [11, 13], the quantisation levels of D-WZ are studied and
optimised to maximise the sum rate under the fixed transmitted
power. However, the distributed compression requires each BS to
have the information about the joint statics of the received signals
across other BSs, and is difficult to be implemented in practice.
Thus, in [10], the authors design simple receiver structures where
a single-user quantiser is deployed on the BSs and successive
interference cancellation (SIC) is applied at the CP. However, all
the works only consider the separated optimisations of transmit
power or quantisation levels and the works about joint
optimisation of transmit power and quantisation levels are very few.
Summary of contributions. In this paper, we focus on the
uplink network MIMO scenario in which each BS is connected
to a CP via a noiseless and rate-limited backhaul link (see Fig. 1).
To implement the compression scheme practically for multicell
processing, the BSs compress the signals using a single-user
quantiser after receiving the signals. Finally, the BSs send the
IET Communications
Research Article
IET Commun., 2015, Vol. 9, Iss. 16, pp. 2041–2047
2041
&
The Institution of Engineering and Technology 2015