IEEE Communications Magazine - June 2017 - page 74

IEEE Communications Magazine • June 2017
72
0163-6804/17/$25.00 © 2017 IEEE
A
bstract
Supporting ultra-reliable and low-latency com-
munications (URLLC) is one of the major goals
in 5G communication systems. Previous studies
focus on ensuring end-to-end delay requirement
by reducing transmission delay and coding delay,
and only consider reliability in data transmission.
However, the reliability reflected by overall pack-
et loss also includes other components such as
queueing delay violation. Moreover, which tools
are appropriate to design radio resource alloca-
tion under constraints on delay, reliability, and
availability is not well understood. As a result,
how to optimize resource allocation for URLLC
is still unclear. In this article, we first discuss the
delay and packet loss components in URLLC and
the network availability for supporting the quali-
ty of service of users. Then we present tools for
resource optimization in URLLC. Last, we summa-
rize the major challenges related to resource man-
agement for URLLC, and perform a case study.
I
ntroduction
Ultra-high reliability (say 10
–7
packet loss proba-
bility) and ultra-low latency (say 1 ms end-to-end
[E2E] delay) are required by a variety of applica-
tions such as autonomous vehicles, factory auto-
mation, virtual and augmented reality, remote
control, and healthcare [1, 2]. As summarized in
[3], ultra-reliable and low-latency communications
(URLLC) lies in the overlapped area of the Inter-
net of Things and tactile Internet, which is one of
the major research directions for fifth generation
(5G) cellular networks [4].
Some technical issues of the network archi-
tecture, wireless access, and resource allocation
for tactile Internet have been discussed in [2, 3],
where E2E delay consists of transmission delay,
coding delay, computing delay, and propagation
delay, and reliability is captured by transmission
error. These studies focus on global communia-
tion scenarios, where the communication distance
ranges from hundreds to thousands of kilome-
ters, and the propagation delay dominates the
E2E delay.
It is worth noting that guarantee the stringent
quality of service (QoS) in terms of both laten-
cy (defined as E2E delay) and reliability (defined
as overall packet loss probability) for URLLC is
not easy even in local communications scenari-
os, where users are associated with a few adja-
cent base stations (BSs) and the communication
distance is less than a few kilometers. In [2],
resource allocation with mixed tactile Internet and
regular traffic was discussed. However, resource
management for URLLC in the radio access net-
work is challenging even if the system only sup-
ports one class of traffic. When designing radio
resource allocation for traditional human-to-hu-
man (H2H) communications, the blocklength
of channel codes is sufficiently large such that
Shannon’s capacity is an accurate approximation
of the achievable rate. However, this is not true
for URLLC, where small packets are transmitted.
Since only a small amount of bits is transmitted
in one coding block and the transmission delay
should be very low, the transmission is not error-
free with finite blocklength channel codes. When
designing resource allocation for URLLC, the sys-
tems need to control the packet loss caused by
transmission error. Therefore, Shannon’s formula
can no longer be applied, because it cannot char-
acterize the maximal achievable rate with given
error probability [5].
Moreover, packet loss may result from factors
other than transmission error, such as queueing
delay violation. Since some event-driven pack-
ets generated by different mobile users (MUs)
arrive at a BS randomly, and the inter-arrival time
between packets may be shorter than the trans-
mission duration of each packet, there is a need
to consider queueing delay [6]. As a result, the
overall packet loss not only comes from uplink
(UL) and downlink (DL) transmission errors, but
also from queueing delay violation. Because E2E
delay and overall reliability are composed of
multiple components, the queueing delay should
be characterized by a delay bound and a delay
bound violation probability for URLLC. Then tools
for analyzing average queueing delay cannot be
used. There are two kinds of tools that have been
applied in analyzing queueing delay of URLLC in
the existing literature. One is network calculus [7],
and the other is effective bandwidth and effec-
tive capacity [8]. However, when these tools are
applicable (and even whether or not they can be
applied) on imposing the constraint on queueing
delay for URLLC is not well-understood.
Different from latency and reliability, which are
Radio Resource Management for
Ultra-Reliable and Low-Latency
Communications
Changyang She, Chenyang Yang, and Tony Q. S. Quek
A
gile
R
adio
R
esource
M
anagement
T
echniques
for
5G N
ew
R
adio
The authors first discuss
the delay and packet loss
components in URLLC
and the network avail-
ability for supporting the
quality of service of users.
Then they present tools
for resource optimization
in URLLC. Last, they sum-
marize the major chal-
lenges related to resource
management for URLLC,
and perform a case study.
Changyang She and Chenyang Yang are with Beihang University; Tony Q. S. Quek is with Singapore University of Technology and Design.
Digital Object Identifier:
10.1109/MCOM.2017.1601092
1...,64,65,66,67,68,69,70,71,72,73 75,76,77,78,79,80,81,82,83,84,...228
Powered by FlippingBook