IEEE Wireless Communications - April 2017 - page 17

IEEE Wireless Communications • April 2017
15
number of accessing devices can cause unaccept-
able delays for the mission critical traffic of DSSE
and demand-response applications, where cer-
tain reliability requirements exist. Since all random
access requests are treated equally in legacy LTE,
there is no way to prioritize certain types of traffic.
For mission critical traffic, we propose an alter-
native approach to random access in LTE, which
makes it possible to reserve sufficient random
access opportunities (RAOs) to ensure a certain
level of success probability (reliability) in the ran-
dom access procedure [11]. Instead of serving all
accessing devices equally, as in the legacy LTE ARP,
the modified approach that is described in detail in
[11] allows the creation of prioritized traffic classes
for which the probability of successfully accessing
the network can be guaranteed. The principle of
this approach is illustrated in Fig. 4, where the priori-
tized traffic classes
m
PMU, PMC/SM, and best effort
(ordered by most important first) that are relevant
for the considered smart grid communication sys-
tem are shown. Each of the
m
PMU and PMC/SM
traffic classes have a dedicated estimation slot in
the contention frame, in which the corresponding
devices must activate a random preamble to access
the network. This enables the eNodeB to estimate
the number of accessing devices and to dimension
the following serving phases to satisfy the required
reliability. Next, the devices will activate a preamble
randomly within the corresponding serving phase, to
start the ARP. Any remaining RAOs in the conten-
tion frame will be available for best-effort traffic. The
duration of the contention frame is set as half of the
shortest latency deadline, to ensure that all deadlines
can be fulfilled.
The allocation of resources for the increasing
number of
m
PMU and PMC/SM devices in an LTE
cell is shown in Fig. 5a. The plot shows that as the
number of active devices increases, first the RAOs
available for best-effort traffic run out. Hereafter,
the RAOs for PMC/SM traffic are sacrificed to
prioritize the more important
m
PMU traffic. Finally,
the total number of devices becomes too large to
also support
m
PMU traffic.
For comparison, in Fig. 5b we show the achiev-
able reliability of the legacy LTE ARP, calculat-
ed using the collision probability model in [12].
Assuming that the arrival of
m
PMU and PMC/SM
traffic occurs in a traditional best-effort manner,
the reliability of all traffic in legacy LTE will drop
below the required reliability of both
m
PMU and
PMC/SM at an earlier point than with our pro-
posed scheme. Notice that the required reliability
of
m
PMU can be supported with the proposed
scheme for three times as many active devices as
with legacy LTE for the R = 1/10 scenario. This
does not hold for the R = 1/3 scenario, where
a larger fraction of devices require 99.9 percent
reliability, since it requires more RAOs to ensure
99.9 percent reliability than 95 percent reliability.
Unfortunately, the proposed scheme for guar-
anteed reliability random access cannot be easily
implemented in today’s cellular networks, since
it requires changes to the LTE protocol in both
devices and eNodeB. However, the scheme
might inspire the development of Machine Type
Communication (MTC) protocols for the upcom-
ing 5G standards.
C
onclusions
and
F
uture
S
teps
With the increasing penetration of distributed
energy resources (DER), the smart grid needs
more and deeper monitoring and control to
maintain stable operation. In this article, we
F
igure
4.
Proposed contention frame layout for
m
PMU, PMC/SM and best effort traffic.
Serving phase
Estimation phase
Estimation
RAO PMU
Serving RAOs for PMU
Serving RAOs for PMC/SM
Serving RAOs
for best effort
Estimation
RAO PMC/SM
F
igure
3.
PMC/SM (dashed line) and
m
PMU (solid line) maximum delay: a) R = 1/10 (moderate DER penetration); b) R = 1/3 (heavy
DER penetration).
Active population
(a)
500
0
10
-1
10
-2
Max delay [s]
10
0
10
1
1000 1500 2000 2500 3000 3500 4000
Active population
(b)
500
0
10
-1
10
-2
Max delay [s]
10
0
10
1
1000 1500 2000 2500 3000 3500 4000
6 PRBs
20 PRBs
50 PRBs
6 PRBs
20 PRBs
50 PRBs
1...,7,8,9,10,11,12,13,14,15,16 18,19,20,21,22,23,24,25,26,27,...132
Powered by FlippingBook