CIRED 2022 Shanghai Workshop Conference 21 – 22 September 2022

Paper 40

Research and Application Test of Communication

Schemes for Differential Protection of Distribution
Network Based on 5G
SHEN Bing1,2, ZOU Xiaofeng1,2*, YAO Jiawei3, CAI Xinchen1,2
1
State Grid Shanghai Municipal Electric Power Company, Shanghai, China
2
East China Electric Power Test and Research Institute, Shanghai, China
3
Shanghai Huawei Technologies Co., Ltd., Shanghai, China
*[email protected]

Keywords: 5G, Differential Protection, Distribution Network, Test

Abstract
As one of the most important application scenarios of 5G, uRLLC(Ultra-Reliable and Low-Latency Communication) provides
end to end latency as low as tens of milliseconds, which could be applied in control and protection systems of Distribution
Network of large cities, such as differential protection. However, simply transplanting the original device into 5G cannot
ensure the optimal operational performance like reliability and speed. Therefore, it is necessary to collect more application
data through field test under variable 5G environment, optimize the differential protection algorithm accordingly. First, the
typical communication schemes of 5G differential protection are compared. Secondly, a 5G test network is built in
professional 5G laboratory to simulate the 5G public network and private network. With three types of 5G CPE(Customer
Premise Equipment), the network tester is used to test 5G end-to-end latency and packet loss rate. Finally, the operation of the
5G differential protections is evaluated, including the behavior of protection under internal and external faults. The test results
above can provide direction for the optimization of communication scheme of 5G differential protection in future.

1 Introduction industry has a long history of applications of advanced

communication technologies. Due to uRLLC, the
As the first full set of 5G standards, Release 15 was characteristics of 5G well meet the needs of
completed by 3GPP(3rd Generation Partnership Project) in communication of control and protection systems in
2017, which mainly focus on eMBB(Enhanced Mobile distribution network of large cities.
Broadband) applications including high definition(HD)
mobile video. In April 2020, Release 16 was completed 5G could replace optical fibre in some applications of city
with enhancement for uRLLC services, providing end to distribution network. However, it is undeniable that 5G is
end(E2E) latency as low as tens of millisecond, which less reliable than traditional optical fibre systems due to
could be applied in industrial control. China has launched wireless communication in access network and resource
5G commercial services since 2019, with leading sharing in transport network and core network of 5G. But
construction speed and coverage of 5G base stations. on the other hand, one single distributed control protection
Telecom operators and industrial specialists are working generally covers only 2 square kilometre area. Therefore,
together to explore 5G use cases in vertical industries, compared with the high voltage power transmission
especially in power grid[1]. network, the influence of unreliability in 5G
communication of distribution network is acceptable. In
On this background, using 5G in communication of control conclusion, 5G is more suitable for protection and
and protection systems in power distribution network has automation systems of medium and low voltage
become a common choice[2-6], which is the integration of distribution network.
traditional relay protection and new telecommunication 3 Analysis of communication schemes of 5G
technology. Differential Protection
As relay devices normally sample 24 points per
2 Application Analysis of 5G in
cycle(20ms), communication schemes of 5G protections
Distribution Network could be classified into two types(Table 1): point by point
From the power line carrier communication, to microwave, packet(mode A) or merged-point packet(mode B). Since
to optical fibre, and to 4G private network, the power 5G uses IP communication rather than MAC,

Authorized licensed use limited to: Kwame Nkrumah Univ of Science and Technology. Downloaded on March 19,2024 at 15:05:34 UTC from IEEE Xplore. Restrictions apply.
 CIRED 2022 Shanghai Workshop Conference 21 – 22 September 2022

Paper 40

manufacturers of 5G protection add layer 3 headers to the

Send Send
original layer 2 packet to realize communication on 5G. Packet 1 Packet 2

Packet 1 Packet 2 Packet 3

(a) Point by Point Packet(Mode A)

For the sending interval of less than 1ms brings high t1 t2

pressure on 5G network, UDP protocol with smaller

packet is generally used instead of TCP in this mode. Fig.1 Delay caused by merged-point packet
Hence, in order to eliminate the influence of packet loss in
high-frequency communication, one sampling point will 4 Main Conditions of Application Test
be sent repeatedly in several packets within a period of
time. Therefore, due to the redundancy method, as long as Test Conditions: A 5G SA(Standalone) test network is
5G is not interrupted completely, a small amount of packet built in the laboratory, and 5G CPE on both sides are under
the same indoor base station. Data exchange between
loss will not affect the operation of differential protection.
internal and external networks is realized through
DMZ(Demilitarized Zone).
(b) Merged-point Packet(Mode B)
5G Network: In terms of access network, two kinds of 5G
Firstly, compared with Mode A, Mode B merges several
base stations are tested. One is the public indoor base
consecutive sampling points into one packet, which could
station(5G Public Network), under which resources are
greatly reduce the frequency of packet sending. Secondly,
shared with other services(such as robot control, video
some manufacturers use TCP protocol with handshake and
surveillance, etc. )during test, bringing interference on the
re-transmission mechanism to avoid packet loss instead of
wireless side. Two wireless resource allocations, with and
packet redundancy method. However, in the case of
without QoS(quality of service), are tested. Another is to
network congestion, the re-transmission triggered by
put the indoor base station in a shielding box (5G Private
packet loss could increase the network pressure of 5G,
Network), simulating hard slicing method in access
which may cause even more packet loss. Thirdly, Mode B
network with 100% resource isolation, that is all resource
may increase delay of relay tripping. Take 8-points packet
blocks(RB) are used in this test. As transport network and
as an example, as shown in Fig. 1, in the most
core network are only for lab test, with data flow much
unfavourable case, when the fault occurs at time t1, it will
less than that of the commercial network. Thus, sharing
not send packet 2 until time t2, which will bring delay of 8
transport network and core network during test should not
sampling intervals(6.67ms) to the tripping of differential
affect the results.
protection.
5G CPE: Typical 5G terminals are tested as in Table 2.
Therefore, Mode B is recommended for 5G differential
protection and self-healing system of distribution network. Table 2: Typical 5G CPE tested
However, to avoid re-transmission, UDP protocol with
packet redundancy should be added. For example, the Type Components Baseband Application
same packet could be sent two or three times consecutively A Chipset Balong 5000 Commercial
to overcome the occasional packet loss of 5G. In the B Module Snapdragon X55 Industrial
application of distribution network, the typical setting C Module Balong 5000 Industrial
value of actuation time of relay is more than hundreds of 5 Wiring and Content of Tests
milliseconds, which has good tolerance for the additional
delay caused by merged-point packet. 5.1 Performance Test of 5G Network

Table 1: Communication Modes of 5G Differential

Protection 5G
network

Mode A B
Packet CPE 1 CPE 2
1200/second 150/second
Frequency
Size of Packet 150 bytes 250 bytes Ethernet
Ethernet
Protocol UDP TCP Port 1 Port 2

packet Network tester

Yes No
redundancy

Authorized licensed use limited to: Kwame Nkrumah Univ of Science and Technology. Downloaded on March 19,2024 at 15:05:34 UTC from IEEE Xplore. Restrictions apply.
 CIRED 2022 Shanghai Workshop Conference 21 – 22 September 2022

Paper 40

Fig.2 Performance Test of 5G Network As in Fig.4, in order to simulate internal and external faults,
relay protection tester applies power signals to differential
In order to test the performance of 5G network, 5G CPE 1 protection 1 and 2, simulating terminals of both sides of
and 2 are connected to the Ethernet Port 1 and 2 of the the power line. The two differential protection devices are
network tester respectively, and packets of Mode A and B connected to GPS synchronous clock to realize
without redundancy in Table 1 are simulated. The same synchronous sampling. Two types of distribution terminals
test is carried out from port 1 to port 2, and port 2 to port 1 from different manufacturers, Type A (Mode A) and Type
at the same time. In order to accurately evaluate network B (Mode B), are tested.
performance, more than 200,000 packets are tested every
time. Thus, test time for Mode A is 5 minutes, and for Additionally, Type A differential protection needs double
Mode B is 25 minutes. The E2E average latency, E2E communications to complete fault identification,
maximum latency, latency distribution and packet loss rate while Type B only needs one-way E2E communication.
in both directions are recorded.
6 Test Results
5.2 Latency Test of Uplink and Downlink
6.1 Comparison of Latency of Mode A and B
Core
Using the test wiring of Fig.2, in order to eliminate the
Network influence of 5G terminal and network, both test cases use
5G CPE A, for its latency performance is more stable
(Table 5), and both carried out in private network to
eliminate interference of other services(Table 4). The test
5G Indoor Station results(Table 3) show that the maximum latency can be
Direct Access
significantly reduced by packing and sending the sampling
CPE 1 points once every eight points(Mode B). According to
typical configurations of telecom operators, one 5G radio
frame is normally 5ms, including 10 slots with 1~2 uplink
Ethernet Ethernet
Port 2
slots. Therefore one uplink cycle is about 5ms, excessively
Port 1
Network tester
frequent packet delivery with period less than 5ms could
only increase the network burden, and thus increase the
maximum latency significantly.(Route 1 represents CPE 1
to 2, and Route 2 represents CPE 2 to 1 in Table 3)

Fig.3 Latency Test of Uplink and Downlink Table 3: Comparison of Latency of Mode A and B

As shown in Fig.3, reconnect port 2 to direct access of Avg Max

core network, therefore CPE 1 to direct access could Test Case Route
Latency Latency
simulate the uplink route of 5G, and direct access to CPE 1 CPE A
could simulate the downlink route. The test items are the 1 6.47ms 92.42ms
Private Network
same with the performance test of 5G network in 5.1. Mode A 2 6.81ms 41.56ms
CPE A 1 6.02ms 15.09ms
5.3 Application Test of 5G Relay Protection Private Network
Mode B 2 6.64ms 16.96ms

6.2 Comparison of Latency of Private and Public

5G network
Networks
The public network(without QoS), public network with
QoS (soft slicing) and private network(hard slicing) are
CPE 1 CPE 2
tested with CPE A and Mode A. The test results(table 4)
GPS
synchronous clock
show that interference from other services greatly
Differential
protection 1
Differential
protection 1 increases the maximum Latency, while QoS of packet has
little effect to improve it. However, the average latency
Relay protection tester and stability of private network with resource isolation are
the best. In addition, the inconsistency of two-way E2E
latency of public network under frequent packet delivery is
obvious, which may be related to routing strategy of core
Fig.4 Application Test of 5G Relay Protection network.

Authorized licensed use limited to: Kwame Nkrumah Univ of Science and Technology. Downloaded on March 19,2024 at 15:05:34 UTC from IEEE Xplore. Restrictions apply.
 CIRED 2022 Shanghai Workshop Conference 21 – 22 September 2022

Paper 40

Table 4: Comparison of Latency of Private and Public Table 6(a): Comparison of Tests of Normal and Longer
Networks Time in Mode A

Avg Max Test Rou Avg Max Package

Test Case Route
Latency Latency Case te Latency Latency Loss
CPE A 1 150.04ms 306.72ms CPE A 1 9.85ms 27.69ms 0
Public Network Private
Mode A 2 14.12ms 120.49ms Network
CPE A Mode A 2 10.55ms 18.0ms 0
Public Network 1 190.41ms 272.14ms 5min
with QoS CPE A
Mode A 2 179.71ms 310.29ms Private 1 11.71ms 130.80ms 0.03‰
CPE A 1 6.47ms 92.42ms Network
Private Mode A 2 10.40ms 67.50ms 0.33‰
Network 2 6.81ms 41.56ms 2hr
Mode A

6.3 Comparison of Latency of different 5G CPE Table 6(b): Comparison of Tests of Normal and Longer
5G CPE A, B, C (Table 2) are tested in private network. Time in Mode B
As CPE has sufficient capability when dealing with the
packets of a single terminal, it is not likely to bring large Test Avg Max Package
Route
occasional latency. Since the maximum latency fluctuates Case Latency Latency Loss
greatly with the network conditions, the average latency is CPE A 1 8.04ms 91.3ms 0
a better indicator of CPE performance. Table 5 shows that Private
although 5G integrated module can reduce the size of CPE Network
and reduce the difficulty of CPE development, the overall Mode B 2 8.91ms 18.80ms 0
performance of CPE based on 5G module is worse than 25min
that based on chips. CPE A 1 8.48ms 90.92ms 0.15‰
Private
Table 5: Comparison of Latency of different 5G CPE Network
Mode B 2 9.18ms 94.75ms 0.10‰
Avg 12hr
Test Case Route
Latency
CPE A 1 6.47ms 6.5 Comparison of Latency of Uplink and Downlink
Private Network As in Fig.3, Mode A is tested in private network and
Mode A 2 6.81ms public network with QoS respectively. The test results
CPE B 1 12.60ms show that the uplink latency is greater than the downlink
Private Network latency, which is mainly due the smaller transmitting
Mode A 2 14.13ms power of 5G CPE compared to than that of the base station,
CPE C 1 10.91ms and the wireless interference caused by multiple services.
Private Network
Mode A 2 7.90ms Table 7: Comparison of Latency of Uplink and Downlink

6.4 Comparison of Tests of Normal and Longer Time Avg Max

Test Case Route
To test 5G performance for longer time, time of Mode A Latency Latency
test is increased to 2 hours (8.64 million packets) and time 5G CPE A Uplink 3.44ms 12.92ms
of Mode B test is increased to 12 hours (6.48 million Private
packets). Table 6 compares tests of normal and longer time Network Downlink 3.91ms 8.44ms
in Mode A and Mode B respectively. The results show that Mode A
the maximum latency increases with the test time in the 5G CPE A Uplink 10.92ms 214.62ms
private network, and a small amount of packet loss occurs, Public
thus the redundancy method is necessary in 5G. Mode B Network
with lower packet frequency performs better in long time with QoS Downlink 4.18ms 8.39ms
operation. Mode A

Authorized licensed use limited to: Kwame Nkrumah Univ of Science and Technology. Downloaded on March 19,2024 at 15:05:34 UTC from IEEE Xplore. Restrictions apply.
 CIRED 2022 Shanghai Workshop Conference 21 – 22 September 2022

Paper 40

6.6 Test of External Faults 5G CPE A

As in Fig.4, the current simulating external faults is Public
applied, which is twice that of the setting value of the Network 46.1ms 16ms/18ms
differential protection. Each fault lasts for 10 seconds, and Differential
is repeated for 25 times. Table 8 shows three test cases Protection B
with no false trip, which indicate that based on 7 Conclusion
synchronous sampling, the differential protection could
function correctly in 5G under external faults. In general, 5G has occasional large latency and packet loss,
but its probability is relatively low. Although the latency
could be increased by merged-point packet(Mode B), it is
Table 8: Tests of External Faults acceptable in the distribution network. Especially self-
healing system, designed to operate after relay protection,
Test Case False Trip has better tolerance for increased latency, which could
5G CPE A greatly improve the power supply reliability of large cities.
Public Network Due to the better performance, Mode B is recommended.
No
with QoS
Differential Protection A The 5G standards are evolving and improving its latency
5G CPE A and reliability. It requires continuous testing for the
Public Network updating hardware and software of terminals and base
No stations, to match the best application scenarios and the
with QoS
Differential Protection B communication schemes.
5G CPE B
Public Network 8 Acknowledgements
No
with QoS
Differential Protection A This paper is sponsored by “Shanghai Science and
Technology Committee Rising-Star
6.7 Test of Internal Faults Program”(20QB1400500).
The current simulating internal faults is applied, which is
twice that of the setting value of the differential protection, 9 References
and the test is repeated 10 times. Table 9 lists the average
[1] China Academy of Information and Communications
trip time and average latency shown in screens of the two
differential protection devices. Since differential protection Technology: '5G use cases for verticals China'. White
A needs twice the E2E latency, which is 37.3ms, thus the Paper, 2020.
total time of internal processing and relay trip is 21.6ms. [2] Huawei Technologies Co., Ltd: '5G opening up new
And differential protection B needs only one E2E latency, business opportunities'. White Paper, 2016.
which is 16ms, thus the total time of internal processing [3] WIKSTRÖM G, TORSNER J, KRONANDER J, et al.:
and relay trip is 30.1ms. 'Wireless protection of power grids over a 5G network'.
2019 IEEE PES GTD Grand International Conference and
Test cases show that the differential protections trip
correctly under internal faults, and the trip time varies with Exposition Asia (GTD Asia), Bangkok, Thailand, 2019,
the 5G latency. Considering the application scenario of pp. 976-981.
distribution network, the typical setting value of trip time [4] CHEN Congjing, HUO Yinlong, YUE Feng, et al.:
is normally over hundreds of milliseconds, trip time of 5G 'Research and application of data synchronization method
differential protection can meet the application for differential protection of distribution network based on
requirements.
5G communication', Zhejiang Electric Power, 2021, 40,
(2), pp. 68-72.
Table 9: Tests of Internal Faults
[5] WANG Changling, ZHAO Yuan.: 'Analysis of latency
Average Average Latency and jitter of power differential protection based on 5G
Test Case bearer network', Communications World, 2019, (32), pp.
Trip Time on Screen
5G CPE A 33-36.
Public [6] WANG Tinghuang, YU Jiang, XU Jian, et al.:
Network 58.9ms 11.4ms/25.9ms 'Discussion on adaptive differential protection technology
Differential of distribution network based on 5G wireless
Protection A
Communication', Distribution & Utilization, 2019, 36, (9),
pp. 18-21.

Authorized licensed use limited to: Kwame Nkrumah Univ of Science and Technology. Downloaded on March 19,2024 at 15:05:34 UTC from IEEE Xplore. Restrictions apply.