Capacity Analysis and Link Scheduling Design in Double-Channel Full-duplex
Wireless Networks
Lulu Li
∗
, Zhuo Li
∗†‡
, Xin Chen
∗
∗
School of Computer Science, Beijing Information Science & Technology University, Beijing 100101, P.R.China
†
Beijing Key Laboratory of Internet Culture and Digital Dissemination Research,
Beijing Information Science & Technology University, Beijing 100101, P.R.China
‡
Corresponding Author. E-mail: lizhuo@bistu.edu.cn
Abstract—Full-duplex communication has been proved as a
realizable way to enhance the capacity of wireless network.
In this paper, we analyze the asymptotic capacity of full-
duplex wireless networks where each terminal is equipped
with only one interface, but two channels are available. By
analysis, we obtain the upperbound of the full-duplex capacity.
Based on the theoretical analysis, we design two link scheduling
algorithms. One is a Centralized Link Scheduling Algorithm
(CLSA) based on Maximal Independent Set (MIS) Selection.
The other is a Distributed Link Scheduling Algorithm (DLSA)
following the idea Carrier Sensing Multiple Access. In CLSA,
we design a simulated annealing based mechanism for MIS
selection problem. In DLSA, we modify the subtraction of
backoff timer when it is non-zero to effectively utilize idle
time in networks. We investigate the performance of these two
algorithms through simulations. For comparison, we implement
a Greedy Link Scheduling Algorithm (GLSA). It is found that
the throughput obtained from CLSA is about 15%-25% higher
than that obtained from GLSA. Simulation results also show
that our proposed DLSA performs almost as good as CLSA
and GLSA and its time complexity and space complexity are
lower than CLSA.
Keywords-full-duplex; capacity analysis; link scheduling;
double-channel;
I. INTRODUCTION
In the traditional wireless network technologies, it is
impossible for a wireless node to transmit and receive signal
simultaneously. This is because that interference generated
by the outgoing signal from itself is much stronger than the
incoming signal, which is referred as self-interference (SI).
SI is basically the loopback of a transmitted signal at the
receiver, whose power is usually about 110dB stronger than
that of the desired signal sent by a far-end transmitter [1].
By using Antenna Cancellation, Jung Il Choi et al. achieve
the amount of SI cancellation required for full-duplex oper-
ation [2] [3]. Full-duplexing become into reality. Compared
with the traditional half-duplex, full-duplex technology can
improve the spectrum efficiency significantly. Driven by
ever-increasing demands for wireless services over a limited
spectrum, full-duplexing attracts extensive attention.
In traditional half-duplex wireless communication, nodes
can use different non-overlapped channels from their neigh-
bors to avoid interference, which is referred to as the multi-
channel technology in [4]. It can improve the network
capacity significantly, especially when multiple radios are
available [5]. Researchers have investigated the multi-radio
multi-channel technology for a long time, which has usually
been used in wireless mesh networks [5]. However, there is
little work on the problem of the performance that can be
obtained when multi-channel allocation is introduced into
full-duplex wireless networking.
In this paper, we consider the scenario where two or-
thogonal channels are available in the full-duplex wireless
networks. The main contributions are as follows:
1) We develop a comprehensive analytical framework to
analyze the asymptotical network capacity of full-duplex
with only one interface and two orthogonal channels and
obtain its upperbound.
2) Based on the analysis of network capacity, we design
two link scheduling algorithms. One is a Centralized Link
Scheduling Algorithm (CLSA) based on Maximal Indepen-
dent Set (MIS) Selection and the other is a Distributed Link
Scheduling Algorithm (DLSA) following the idea Carrier
Sensing Multiple Access (CSMA). In CLSA, we design a
simulated annealing based algorithm for the MIS selection
problem. In DLSA, we modify the subtraction of backoff
timer when it is non-zero to effectively utilize idle time
in networks. The time complexity and space complexity of
CLSA are both 𝑂(𝑁
2
), where 𝑁 is the number of nodes in
wireless networks.
3) We conduct simulations on MATLAB to investigate
the performance of CLSA and DLSA. For comparison, we
also implement two other link scheduling schemes, i.e.,
ideal link scheduling method and Greedy Link Scheduling
Algorithm (GLSA). It is found that the throughput obtained
from CLSA can approach to the theoretical throughput and
achieve about 15%-25% higher than that obtained from
GLSA. And our proposed DLSA performs not so good as
CLSA. It is acceptable because that there is asynchronous
contention in CSMA. More over, there are some advantages
of DLSA, e.g., the lower time complexity, space complexity
and its easiness to implement.
The rest of this paper is organized as follows. We discuss
the related work in Sec. II and describe the system models
2016 IEEE TrustCom/BigDataSE/ISPA
2324-9013/16 $31.00 © 2016 IEEE
DOI 10.1109/TrustCom/BigDataSE/ISPA.2016.243
1582
2016 IEEE TrustCom/BigDataSE/ISPA
2324-9013/16 $31.00 © 2016 IEEE
DOI 10.1109/TrustCom/BigDataSE/ISPA.2016.243
1582
2016 IEEE TrustCom/BigDataSE/ISPA
2324-9013/16 $31.00 © 2016 IEEE
DOI 10.1109/TrustCom/BigDataSE/ISPA.2016.243
1582
2016 IEEE TrustCom/BigDataSE/ISPA
2324-9013/16 $31.00 © 2016 IEEE
DOI 10.1109/TrustCom/BigDataSE/ISPA.2016.243
1582
2016 IEEE TrustCom/BigDataSE/ISPA
2324-9013/16 $31.00 © 2016 IEEE
DOI 10.1109/TrustCom/BigDataSE/ISPA.2016.243
1581