High-Performance Wireless Communications for Smart Grids
KONSTANTINA ROUSSOU
KTH ROYAL INSTITUTE OF TECHNOLOGY
SCHOOL OF ELECTRICAL ENGINEERING AND COMPUTER SCIENCE
Communications for Smart Grids
KONSTANTINA ROUSSOU
Master in Information and Network Engineering Date: November 10, 2020
Supervisor: Michele Luvisotto Examiner: Carlo Fischione
School of Electrical Engineering and Computer Science Host company: Hitachi ABB Power Grids
Swedish title: Högpresterande Trådlös Kommunikation för Smarta Elnät
Abstract
The increased energy needs of today’s society, as well as the necessity for im- proved and more sustainable energy management impose the transition from the traditional power grids to the smart grids. Smart grids are enabled by the advances in Information and Communication technologies and automation.
Such an implementation would allow more efficient energy management and distribution, as well as increase the reliability and its capability to host dis- tributed renewable sources, as required to meet decarbonization goals.
Wireless communications are expected to take an active part in the realiza- tion of smart grids, since they reduce the installation cost and allow deploy- ment flexibility. The choice between wired or wireless communications for each communication link is made depending on the application requirements, as well as the environment around the link. Even though wireless communi- cations are more cost efficient and easy to deploy, wired communications are usually more reliable. However, the benefits gained from the adoption of wire- less communications concern both the energy providers and the end users, and therefore are very significant. Hence, a lot of research has been conducted on several wireless technologies for smart grid applications. This thesis investi- gates a novel wireless technology, the IEEE 802.11ax WiFi standard, and its potential applicability for smart grid communications, and more specifically, for substation communications.
The performance of 802.11ax was tested through simulations. The net- work performance was in general sufficient as long as the number of WiFi clients was limited (up to 7), while progressively decreasing with a higher number of clients. OFDMA is a multiple access technology introduced in 802.11ax that was not tested in this thesis, and is worth looking into, as it could potentially satisfy the imposed requirements.
Sammanfattning
Det ökande behovet av energi från dagens samhälle tillsammans med nödvän- digheten av hållbarhet bidrar till en övergång från traditionellt kraftnät till det smarta elnätet. Det smarta elnätet utvecklas i samband med framgångar inom information- och kommunikationsteknik samt automation. En övergång till ett smart kraftnät skulle medföra mer effektiv energibehandling och distribution, öka pålitligheten samt möjligheten att ha flera spridda förnybara energikällor, något som krävs för att uppnå målet om ett fossilfritt samhälle.
Trådlös kommunikation förväntas ta en aktiv del i denna övergång, ef- tersom det minskar installationskostnader och tillåter flexibel distribution. Va- let mellan trådlös och trådbunden kommunikation görs beroende på applika- tionskraven samt omgivningen runt länken. Fastän trådlös kommunikation of- tast är billigare och enklare att installera, är trådbunden kommunikation van- ligtvis mer pålitligt. På grund av fördelarna med trådlös kommunikation har mycket forskning gjorts inom många olika teknologier inom området. Denna avhandling fokuserar på en ny standard inom trådlös teknik, IEE 802.11ax Wi- Fi, samt möjligheten att applicera denna standard till det smarta elnätet, mer specifikt för så kallade substations.
Prestandan av 802.11ax testades genom simuleringar. Generellt sett så var standarden tillräckligt, så länge som antalet WiFi klienter var begränsat (upp till 7). Med ökade antal klienter så minskade prestandan av standarden. OF- DMA är en teknologi introducerat i 802.11ax som innefattar flera åtkomst- punkter. Denna teknologi var inte undersökt i denna avhandling, men kan vara intressant för framtida studier då den kanske kan uppfylla de påtvingade kra- ven.
Acknowledgement
I would like to thank my supervisor, Michele Luvisotto, for his valuable guid- ance and support throughout this project. It has been a truly interesting project, through which I learned a lot of new concepts, as well as improved my under- standing of others. I would also like to express my gratitude to Hitachi ABB Power Grids, as well as professor Fischione, for their help and support through- out this endeavor.
Finally, I would like to thank my family and each of my friends for always standing next to me and supporting all of my efforts.
1 Introduction 1
1.1 Background . . . 1
1.1.1 Smart Grid . . . 1
1.1.2 Smart Grid Communications . . . 2
1.1.3 Substation Automation and the utilization of IEC 61850 4 1.2 Related Work . . . 8
1.2.1 Wireless Communications for IEC 61850 . . . 8
1.2.2 WirelessHART . . . 9
1.2.3 Wireless networks for Industrial Automation-Process Automation (WIA-PA) . . . 10
1.2.4 IEEE 802.11ac . . . 11
1.2.5 4G LTE . . . 12
1.2.6 5G . . . 13
1.3 Research Questions . . . 14
1.4 Contribution . . . 14
1.5 Methodology . . . 15
1.6 Report Outline . . . 15
2 Literature Study 17 2.1 The novel IEEE 80211ax standard . . . 17
2.1.1 PHY Enhancements . . . 18
2.1.2 Channel Access and Multi-user (MU) Transmission . . 20
2.1.3 Spatial Reuse and Overlapped BSS (OBSS) manage- ment . . . 23
3 Simulation Work 25 3.1 The Network Simulator NS-3 . . . 25
3.2 The Simulation Scenario . . . 26
3.2.1 Deployment Scenario . . . 26
vi
3.2.2 Traffic Model and WiFi Settings . . . 28
3.2.3 MCS selection . . . 30
3.2.4 Key Performance Indicators . . . 32
3.3 Simulation Results . . . 32
3.3.1 Results for the Large Substation case . . . 32
3.3.2 Results for the Small Substation case . . . 37
3.3.3 Results for the Large Substation using MIMO . . . 40
4 Conclusions 45
5 Future Work 47
Bibliography 49
A Additional Results 51
1.1 Global consumers’ concerns. [1] . . . 2
1.2 The NIST Conceptual Model for SG. [2] . . . 3
1.3 IEC 61850 data modeling. [5] . . . 5
1.4 Message communication stack of IEC 61850. [6] . . . 7
1.5 IEC 61850 communication architecture diagram. [7] . . . 8
1.6 WirelessHART network representation [7]. . . 10
1.7 WIA-PA system architecture [9]. . . 11
2.1 The two parts of the preamble, the legacy and HE part, that are duplicated every 20 MHz [16]. . . 19
2.2 Repetition mode for HE-SIG-A [16]. . . 20
2.3 An UL OFDMA transmission [13]. . . 22
3.1 The simulated deployment scenario. . . 27
3.2 The outdoor 400kV substation scenario [19]. . . 28
3.3 PER vs SNR for several MCS values (64 bytes frame). . . 31
3.4 Mean value of latency vs number of clients for a large substation. 33 3.5 Mean value of PER vs number of clients for a large substation. 34 3.6 P(latency > 3 ms) vs number of clients for a large substation. . 35
3.7 Mean value of latency vs Aggregation number n, for 5 clients for a large substation. . . 36
3.8 Mean value of PER vs Aggregation number n, for 5 clients for a large substation. . . 37
3.9 P(latency > 3 ms) vs Aggregation number n, for 5 clients, for a large substation. . . 38
3.10 Mean value of latency vs Aggregation number n, for 10 clients, for a large substation. . . 39
3.11 PER vs Aggregation number n, for 10 clients, for a large sub- station. . . 40 3.12 Mean value of latency vs number of clients for a small substation. 41
viii
3.13 PER of latency vs number of clients for a small substation. . . 42 3.14 Mean value of latency vs number of clients for a large substa-
tion with the use of MIMO 2x2 (no aggregation). . . 43 3.15 Mean value of latency vs number of clients for a large substa-
tion with the use of MIMO 2x2 (with aggregation). . . 44 A.1 E[P(lat > 3 ms)] vs number of clients for a small substation. . . 51
1.1 Latency Requirements set by IEC 61850 [7]. . . 7 1.2 Technologies that have been studied. . . 9 3.1 Simulation Parameters. . . 30 3.2 Latency statistics comparison according to aggregation value
n, for 5 clients at the Large substation case, at 40 MHz. . . 35 3.3 Latency statistics comparison according to aggregation value
n, for 7 clients at the Large substation case, at 40 MHz. . . 36 3.4 Latency statistics comparison according to aggregation value
n, for 10 clients at the Large substation case, at 40 MHz. . . . 37 3.5 Latency statistics comparison according to aggregation value
n, for 5 clients at the Small substation case, at 40 MHz. . . 39 3.6 Latency statistics comparison according to aggregation value
n, for 10 clients at the Small substation case, at 40 MHz. . . . 40 A.1 Latency statistics comparison according to aggregation value
n, for 7 clients at the Small substation case, at 40 MHz. . . 52
x
3G Third Generation Technology Standard for broadband cellular networks.
12
4G Fourth Generation Technology Standard for broadband cellular networks.
9, 12, 13
5G Fifth Generation Technology Standard for broadband cellular networks.
9, 13
AC Access Categories. 22
AP Access Point. 18, 21, 22, 26, 47 BA Block ACK. 22
BAR Block ACK Request. 22 BS Base Station. 13
BSS Basic service set. 19, 23, 24 CT Current Transformer. 4 CTS ClearToSend. 18
DBCA Dynamic Bandwidth Channel Access. 18 DCM Dual Carrier Modulation. 20
DL Downlink. 19, 20, 21
DSC Dynamic Sensitivity Control. 18 eMBB Enhanced Mobile Broadband. 13
xi
FEC Forward Error Correction. 11 GI Guard Interval. 18, 29, 30
GOOSE Generic Object Oriented Substation Event. 6, 7, 9, 12 GSE Generic Substation Event. 6
HE High Efficiency. viii, 19 HEW High Efficiency WLAN. 17 HMI Human-Machine Interface. 7
HT PHY High-Throughput Physical layer. 12
ICT Information and Communication Technologies. 2 IED Intelligent Electronic Device. 5, 28, 47
IEEE-SA IEEE Standards Association. 17 IP Internet Protocol. 6, 12, 29
KPI Key Performance Indicator. 32 LN Logical Node. 6
LoS Line of Sight. 26, 27, 45
LTE Long Term Evolution. 9, 12, 13 LTF Long Training Field. 12
M2M Machine-to-Machine. 8 MAC Medium Access Control. 23
MCS Modulation and Coding Scheme. vii, viii, 19, 21, 29, 30, 31, 41 MIMO Multiple Input and Multiple Output. vii, ix, 11, 12, 17, 19, 29, 30,
40, 41, 43, 44, 46
mMIMO Massive MIMO. 13
MMS Manufacturing Message Specification. 7, 9 mMTC massive Machine Type Communications. 13 mmWave Milimeter Wave. 13
MU Multi-user. vi, 20, 21, 22
MU-RTS/CTS MU-RequestToSend/ClearToSend. 22 MU-MIMO Multi-user MIMO. 21, 22
NIST National Institute of Standards and Technology. viii, 1, 3 NLoS Non Line of Sight. 26, 27, 45
NS-2 Network Simulator 2. 25
NS-3 Network Simulator 3. 15, 25, 26, 31, 47 NSTS Number of Spatial Streams. 19, 21 OBO OFDMA Back-off. 23
OBSS Overlapped BSS. vi, 23
OFDM Orthogonal Frequency-Division Multiplexing. 18, 19, 20, 30
OFDMA Orthogonal Frequency Division Multiple Access. viii, 15, 18, 20, 21, 22, 23, 47
PER Packet Error Rate. viii, ix, 20, 30, 31, 32, 34, 35, 36, 37, 39, 40, 42, 46, 52
PIU Process Interface Unit. 6 PLE Path Loss Exponent. 27, 40 PT Power Transformer. 4
QoS Quality of Service. 4 RTS RequestToSend. 18 RU Resource Unit. 21, 23, 47
SAS Substation Automation System. 4, 5
SCADA Supervisory Control and Data Acquisition system. 7 SG Smart Grid. viii, 1, 2, 3, 4, 6, 8
SNR Signal-to-noise ratio. viii, 19, 31, 38 STA Station. 18, 19, 20, 21, 22, 23, 26 SU Single-user. 20
SV Sampled Values. 6, 7, 9, 12, 13, 14, 26, 29, 30 TCP Transmission Control Protocol. 6, 25, 29 TDMA Time Devision Multiple Access. 9, 10 UDP User Datagram Protocol. 6, 10, 12 UE User Equipment. 13
UL Uplink. viii, 19, 21, 22 UL MU Uplink Multi-user. 20
URLLC Ultra-Reliable and Low Latency Communications. 13 VHT PHY Very-High Throughput Physical layer. 12
VT Voltage Transformer. 4 WH WirelessHART. 9, 10
WIA-PA Wireless Networks for Industrial Automation - Process Automation.
viii, 9, 10, 11
Introduction
1.1 Background
1.1.1 Smart Grid
The term Smart Grid (SG) refers to an innovative technology infrastructure that is expected to meet the energy requirements of the 21st century, by inte- grating the latest digital communications and advanced control technologies to the existing power grid [1]. According to International Energy Agency, smart grids are defined as "electricity networks that use digital and other advanced technologies to monitor and manage the transport of electricity from all gen- eration sources to meet the varying electricity demand of end-users". Addi- tionally, according to National Institute of Standards and Technology (NIST), some of the anticipated benefits and requirements of the SG would be [2]:
1. Improving power reliability and quality
2. Optimizing facility utilization and averting construction of back-up (peak load) power plants
3. Enhancing capacity and efficiency of existing electric power networks 4. Improving resilience to disruption
5. Enabling predictive maintenance and self-healing responses to system disturbances
6. Facilitating expanded deployment of renewable energy sources 7. Accommodating distributed power sources
1
Consumers are in need of reliable energy supply that is also greener. Cost efficiency and high power quality are important as well. Consumers’ concerns regarding power supply are displayed at figure 1.1.
Figure 1.1: Global consumers’ concerns. [1]
For this transition, the key enablers have been advances in the fields of control theory as well as in the field of Information and Communication Tech- nologies (ICT). Smart grids need to be able to deliver power in an efficient manner and also adapt to a wide range of conditions and events. Those events could occur at any level of the grid, such as power generation, transmission, distribution and consumption, and then actions need to be initiated, according to control outputs [2]. The detection of events can be achieved with the use of smart meters, sensors, data management systems and monitoring systems that control the flow of information among various stakeholders.
According to NIST, the SG can be understood using the Conceptual Model in figure 1.2. This model divides the SG into seven domains: bulk gener- ation, transmission, distribution, consumer, markets, operations and service providers. Each domain is comprised of devices, programs or systems that make decisions and exchange necessary information [2].
1.1.2 Smart Grid Communications
Given the great degree of integration of advanced technologies and applica- tions that will need to be run across all domains of the SG, a huge amount of data is expected to be generated. These data will be subjected to further anal- ysis, control, real-time pricing methods, while also trigger feedback actions
Figure 1.2: The NIST Conceptual Model for SG. [2]
in the grid. All these data transactions will be done over a communication system, and that is why focusing on this communication system implementa- tion is crucial. Although it is clear that the use of communication networks is necessary, the type of communications has not been clarified yet. That can be attributed to the fact that the communication scenarios that will need to be supported are various, including [2]:
• Enterprise buses that connects control center applications, markets, and generators
• Wide area networks that connect geographically distant sites
• Field area networks that connect devices, such as intelligent electronic devices that control circuit breakers and transformers
So far, it is understood that several technologies need to be used throughout the SG, while there is need for certain requirements to be set, so that the op-
timal communication system can be chosen in each case. These requirements include for example [2]:
• Quality of Service (QoS): data which are of critical importance for the overall performance of the SG, needs to be delivered timely
• High reliability: communication networks must provide reliable exchange of information due to the high criticality of SG data
• High availability and coverage: the SG has to be in position to respond in time to any event that occurs anywhere in the network
• Security and privacy guarantee: The protection system of the SG should not only concern inadvertent compromises of the grid (equipment fail- ure, natural disasters), but also cyber attacks
It is well understood that wired and wireless communications are both needed to ensure the information flow in an SG, while choosing between the two depends on each case. Wireless networks are more cost efficient to in- stall, while offering ease of coverage for sites that are difficult to access. At the same time, the wireless medium is more prone to interference and losses [3]. On the other hand, wired communications are much less subject to losses and interference in the medium. However, the installation and maintenance of such networks are more cumbersome and costly to implement [3]. Hence, the type of communication chosen in each case depends on the nature of the site, on the application requirements and on cost considerations.
In this project, the main focus is to investigate whether the use of wire- less communications, and more specifically, of the novel IEEE 802.11ax WiFi standard, could sufficiently cover the needs of SG field area networks for sub- station communications. Therefore, the following sections aim to introduce the key concepts of substation communications, starting from the IEC 61850 standard, that is commonly used in this context.
1.1.3 Substation Automation and the utilization of IEC 61850
According to [4], the primary plant of a substation consists of high-voltage equipment, including circuit breakers, Current Transformers (CTs), Voltage Transformers (VTs) and Power Transformers (PTs). A Substation Automation
System (SAS) covers protection, automation, control, monitoring and meter- ing functions, related to the primary equipment. The links between the pri- mary plant and the SAS are called "process connections" and they are the main focus of this project. Typically, these process connections transfer in- formation (such as alarms, indications and transduced analog data) from the primary plant to the SAS, while another kind of information (operating com- mands and configuration changes) is sent from the SAS to the primary plant.
One feature of this information transfer is the need to maintain interoperability between equipment from several vendors.
The IEC 61850 standard has been identified as a key step towards inter- operability in automation and protection for substation communications. The objective of this standard is to support the existing needs of power utility au- tomation, while also cover future developments as technology improves [4].
The purpose of this standard is to solve the interoperability issues between In- telligent Electronic Device (IED), that originate from different manufacturers.
IEC 61850 defines a hierarchy which considers at the bottom the Logical De- vices, and on top of them Logical Nodes, Data Objects and Data Attributes [5]. This hierarchy can be viewed in figure 1.3.
Figure 1.3: IEC 61850 data modeling. [5]
The IEC 61850 series of standards depicts a SAS as an object-oriented data
model. In this context, the Logical Node (LN) is the smallest entity that can participate in a communication exchange, and LNs are defined in IEC 61850- 7-4. The several functionalities that occur in the SG, are executed by one or multiple LNs. If the LNs are not installed in the same physical device, then communication links between them are required [4].
With the aim to achieve the required performance for each application, IEC 61850 defines seven types of messages [6]:
• Generic Object Oriented Substation Event (GOOSE) and Generic Sub- station Event (GSE) (type 1)
• Raw data Sampled Values (SV) (type 4)
• Time synchronization messages (type 6)
• Medium and low speed messages (types 2, 3, 5, and 7)
The SV and GOOSE/GSE messages (type 1 and 4) are generated from time critical applications, and hence they are directly mapped to the link layer, with the aim to make those transmissions more robust. The medium and low speed messages (types 2, 3, 5, and 7) are designed as client/server Internet Protocol (IP) messages, so they go through the Transmission Control Protocol/Internet Protocol (TCP/IP) stack as well. Regarding the time synchronization messages (type 6), User Datagram Protocol/Internet Protocol (UDP/IP) is utilized. The communication stack for the aforementioned cases is depicted at figure 1.4.
Furthermore, IEC 61850 sets certain latency requirements for each type of messages, according to the nature of the data that is sent. Low latency is of high importance since it is directly related to safety and control efficiency of the grid. Table 1.1 illustrates the latency requirements as set for IEC 61850.
In this case, P1 corresponds to the distribution substations, P2, transmission substations and P3 transmission substations with timing and circuit breaker.
Substation communications can be distinguished in three layers, as shown in figure 1.5. The substation devices can be classified in three levels: sta- tion, bay and process. Looking at them from bottom to top, the process level devices consist of Instrumental Transformers and Intelligent Switchgear Inter- faces, which can be viewed as one Process Interface Unit (PIU). Then the level in the middle is they bay level, which includes metering, control and protec- tion IEDs. Finally, station level is the highest level, which is responsible for
Figure 1.4: Message communication stack of IEC 61850. [6]
the supervision and control of the station, containing interfaces like Super- visory Control and Data Acquisition system (SCADA) and Human-Machine Interface (HMI). The commnication between those three levels is managed by two communication buses. The first one is the process bus that connects the devices from the process level to the bay level. Additionally, there is the station bus, which connects the bay level to the station level [7].
IEC 61850 Latency requirements
Type Class Example Message P1 (ms) P2 (ms) P3 (ms)
1A Fast Messages Trips GOOSE 10 3 3
1B Fast Messages (others) Commands and Simple messages
GOOSE 100 20 20
2 Medium Speed Mes- sages
Measured Values MMS 100 100 100
3 Low Speed Messages Parameters MMS 500 500 500
4 Raw Data Messages Output Data from Digitis- ing Transducers
SV 10 3 3
5 File Transfer Files MMS 100 100 100
6 Time Synchronization Process Bus and Station Bus Clock Synchroniza- tion
TimeSync No requirements (only preci- sion)
7 Command Messages HMI Commands MMS 500 500 500
Table 1.1: Latency Requirements set by IEC 61850 [7].
Figure 1.5: IEC 61850 communication architecture diagram. [7]
1.2 Related Work
1.2.1 Wireless Communications for IEC 61850
Smart Grid (SG) and substation automation are part of the big Industry 4.0 trend, according to which traditional industrial process becomes more con- nected and automated to improve resource efficiency and adaptability. Machine- to-Machine (M2M) communications are an essential part in this process [8]
and IEC 61850 can be classified as a standard for M2M communication in substations.
Most of the M2M communication systems deployed in the field have been implemented using wired communications, while wireless communications have had a more complementary role [8]. However, together with the rise of Industry 4.0, wireless communications have seen an increase in momentum.
These technologies offer increased scalability and simple deployment. They are also well-suited for remote locations or or locations that are hard to reach.
Following this trend, it is worth to investigate the application of wireless com- munication to substation automation scenarios.
Some of the most popular wireless technologies that have been researched for the purpose of substation communications will be presented in the follow- ing subsections. A summary of the technologies and use cases is provided in Table 1.2.
Related Works
Paper Wireless Technology IEC 61850 Traffic Evaluation Method used [7] WirelessHART (WH) GOOSE and MMS Simulation
[9] WIA-PA MMS Application testing in pilot
substation
[10] 8021.11ac GOOSE Measurements from experi-
mental system set-up
[11] 4G LTE SV Lab experiment
[11] 5G SV Simulation
Table 1.2: Technologies that have been studied.
1.2.2 WirelessHART
WirelessHART (WH) is the first certified standard for wireless communica- tions that are aimed for industrial applications by IEC, and it was released in 2007 [7]. The protocol supports operation in the 2.4 GHz ISM band and is based on the IEEE 802.15.4 standard. It supports mesh and star topologies, depending on the distance between the nodes or the number of devices. It is Time Devision Multiple Access (TDMA) based, while also utilizing frequency hopping technologies in order to increase reliability and security in varied en- vironments. The functionality of the network is coordinated by the Network Manager, the Security Manager and access point, that are usually placed in the same box. The network may also consist of other components, such as field devices, that perform basic sensing and actuation in the field, adapters that al- low legacy equipment to be combined with wireless communications, as well as portable devices that are used for calibration. An example of a WH network can be viewed in figure 1.6.
In [7], simulation work has been conducted with the purpose to measure the performance of a system that combines the use of IEC 61850 with WH.
The implemented gateway is converting WH messages to IEC 61850 GOOSE and MMS messages. The network is divided into three parts: the IEC 61850
Figure 1.6: WirelessHART network representation [7].
network (with delay D1), the part between the developed gateway and the WH gateway (with delay D2), and the WH network (with delay D3). The average D1 has been calculated as 0.1 ms, when using 200 MB/s LAN speed. For the WH network, considering the scenario with two hops until the final destination of the message, the average delay of D3 is 1297 ms [7]. Finally, D2 was found to be 2.73 ms on average. In this case, the HART command 3 was sent to the WH Gateway via UDP messages. The total average delay for D1, D2 and D3 was calculated as 1002.83 ms, which is significantly higher than the IEC 61850 requirements, as set in Table 1.1. The high delay is attributed to the fact that WH uses 10 ms time slots in its TDMA implementation.
1.2.3 Wireless networks for Industrial Automation-Process Automation (WIA-PA)
In [9], an implementation of a WIA-PA for substation communication is pre- sented. In this implementation, the equipment in the process level, that needs to be monitored, has integrated wireless adapters, which are used to to trans- mit monitoring data to the wireless intelligent gateway, which is in the bay level. This communication happens over a WIA-PA channel, as it can be seen in figure 1.7. The bay level consists of the wireless intelligent gateway and an IED device. The wireless intelligent gateway sends the data that it received over the WIA-PA interface to the IED device, after converting it into the IEC
61850 standard. The IED device will then send the data to the Station Applica- tion server in the substation level, and will also receive the control instructions as feedback from the Station Application Server.
Figure 1.7: WIA-PA system architecture [9].
The aforementioned system was implemented and tested for a few months in a pilot substation in Liaoning Province, China. The authors claim that WIA- PA can offer great performance in the system and is a promising technology for substation application [9]. However, no detailed measurement result that can back up this claim is reported.
1.2.4 IEEE 802.11ac
The IEEE 802.11ac standard was released in 2014 and introduces many ad- vances to the WiFi physical layer, which significantly improve the performance, in comparison to previous versions of the standard [10]. These improvements include larger aggregated bandwidth, up to 160 MHz for a single transmission, denser modulation of 256-QAM and more efficient Forward Error Correction (FEC) with the use of 5/6 code rate, and additionally the use of Multiple Input and Multiple Output (MIMO) with up to eight spatial streams. These features
are defined as the Very-High Throughput Physical layer (VHT PHY), which can reach data-rates of up to 866.6 Mb/s per spatial stream. The performance can be improved even more when considering the use of MIMO, which can reach up to 6.77 Gb/s. This consists a more than tenfold performance increase when comparing to the 600 Mb/s of total aggregate throughput, which was offered in the previous High-Throughput Physical layer (HT PHY).
The VHT PHY incorporates and improves several features of the HT PHY, such as the sounding options with the associated Long Training Fields (LTF), as well as beam-forming mechanisms. The VHT PHY includes advanced mechanisms for calibrating the transmission power, as well as the capability to dynamically switch from Short to Long Guard Interval, so as to deal with soft or hard multi-path problems, while decreasing the overhead [10].
In [10], the results of a measurement setup are presented. In this setup, IEC 61850 GOOSE messages are sent over 802.11ac network, along with UDP traffic. The performance of the network is tested for several UDP traffic inten- sities. When considering UDP traffic of up to 30 Mbps, the latency is up to 3 ms, satisfying the IEC 61850 requirements. However, for the next tested value of 80 Mbps of UDP traffic, the latency is above the 3 ms threshold.
1.2.5 4G LTE
The Fourth Generation Technology Standard for broadband cellular networks (4G) (also called Long Term Evolution (LTE)) is a wireless cellular technol- ogy. It was first launched in 2010 and offered significant improvement in com- parison to its predecessor, the Third Generation (3G) reaching a peak down- load speed of 1 Gbps and a peak upload speed of 500 Mbps. 4G LTE achieves this speed increase with technologies such as OFDM and MIMO. Although cellular networks are typically used for consumer applications (e.g. internet browsing, social media, video streaming, etc,), there is an increasing trend of using them for industrial applications, pushed by the high offered performance.
An example of a 4G LTE network, based on LTE Release 13, used to trans- mit IEC 61850 SVs is shown in [11], where an experimental testbed was set up in a lab environment. The SVs were sent as UDP datagrams, that were later encapsulated in IP packets and sent over the 4G network. The transmis- sion process consisted of two wireless transmissions, one uplink transmission
from an LTE modem to an LTE base station, and a downlink transmission from the base station to another LTE modem in the same radio cell. The parame- ters that were measured included the latency, the packet loss and the network throughput. The average measured latency varied from 6 to 12 ms, with peaks up to 21 ms, thus not satisfying the IEC 61850 requirements.
Furthermore, in [11], the theoretical transmission latency was calculated when considering LTE Release 15 features, such as mini-slots and semi-persistent.
In this case the transmission time was found to be 0.86 ms, and 1.4 ms for each re-transmission. This means that there is room for only 2 re-transmissions, which is deemed not enough to cover the reliability requirements of the appli- cation.
1.2.6 5G
The Fifth Generation Technology Standard for broadband cellular networks (5G) is the most recent cellular technology, whose first version (Release 15) was launched in 2019. It aims to offer even higher data rates than its predeces- sor, 4G, reaching a peak download data rate of 20 Gbps and a peak upload data rate of 10 Gbps [12] in Enhanced Mobile Broadband (eMBB) applications. It is also meant to enable other applications, such as Ultra-Reliable and Low Latency Communications (URLLC) and massive Machine Type Communi- cations (mMTC). Some key enablers for 5G are the adoption of the mmWave technology, the deployment of small cells, as well as the utilization of Massive MIMO (mMIMO) and beamforming.
In [11], a 5G URLLC network was simulated, where the considered sce- nario consists of a 100m by 125m substation environment. All the User Equip- ments (UEs) are placed inside the switchyard, while the BS is placed in the substation control room. The considered traffic is uplink SV transmission, while there is an additional eMBB macro network surrounding the substation.
The load of the interfering eMBB network varies from 0 to 100%. Additional blockage is also considered inside the URLLC cell, so as to capture the effects of obstacles inside the substation.
The results show that coverage of 99.5% is achieved, even when consid- ering the highest load value of 0.9 for the eMBB network and a lower SINR value of 23 dBm. Thus, [11] concludes that 5G can be successfully deployed for local area substation communications, and could substitute or complement
the preexisting wired technology.
1.3 Research Questions
The electrical power grids of today are constantly converging towards the re- alization of a smart grid. The smart grid will offer significant increase in au- tomation and control in all parts of the grid, as well as increased efficiency in the way electrical power is distributed and managed. Furthermore, power is expected to be handled in a more sustainable and efficient manner. A key step towards this goal is the deployment of high-performance communication networks along the grid. Wireless communications are particularly appealing due to the savings on material and installation costs. As new wireless tech- nologies are emerging, promising to deliver higher performances, it is worth studying them to evaluate whether they could be used for smart grid applica- tions, and also potentially suggest specific improvements to address some of the most critical requirements.
The research questions that are aimed to be answered through this thesis are the following two:
• What is the performance of the use of IEEE 802.11ax standard for se- lected use cases within smart grid communications?
• Can the IEEE 802.11ax be used for the transmission of IEC 61850 Sam- pled Values (SV) traffic from the process level to the bay level of the substation?
1.4 Contribution
This thesis will aim to evaluate whether the novel WiFi standard IEEE 802.11ax is well-suited for smart grid substation communications. More specifically, it will test whether this technology can efficiently cover the latency, reliability and availability requirements that are imposed for the transmission of Sampled Values from the process level towards the bay area of the smart grid substation.
This evaluation will be conducted through network simulations, as well as the- oretical calculations of some aspects of the system. The results will provide an insight for the suitability of 802.11ax for this specific use case.
1.5 Methodology
The main focus of this thesis project is to evaluate whether the use of IEEE 802.11ax WiFi standard would satisfy the performance requirements posed for smart grid substation communications. Therefore it has an exploratory nature.
Before conducting the exploratory part of the project, a literature study took place, during which the nature of the substation communications and the cor- responding requirements were studied. From that, it was understood that the most critical aspects for this scenario are: latency, availability, reliability, secu- rity and installation and maintenance costs. Additionally, a literature survey on the novel IEEE 802.11ax standard was also conducted in order to better com- prehend the new features that have been added and how these could be ben- eficial for substation communications. Key new features that were identified were the shorter Guard Interval options, the higher 1024-QAM modulation scheme and the Orthogonal Frequency Division Multiple Access (OFDMA) multiple access scheme. These features appeared promising in the attempt to implement low-latency, reliable and cost efficient communications for the sub- station scenario.
In order to test the validity of the hypothesis, a simulation set-up was im- plemented, using the tool Network Simulator 3 (NS-3), which is a tool widely used for such purposes. Simulations were repeated 10 times for each unique set of parameter values, and later the collected output was averaged so that the accuracy of the results presented is higher. The statistics that were generated are: mean value of latency, standard deviation of latency, probability of error and number of consecutive lost packets. These values were computed using MATLAB.
Finally, concluding remarks were presented, explaining whether the orig- inal hypothesis for the use of IEEE 802.11ax for substation communications was eventually proven to hold, according to the simulated results.
1.6 Report Outline
The current chapter was an introduction to the smart grids and the need for efficient communications for smart grid applications. An overview of the IEC 61850 standard was also presented, so as to state the significance of this stan-
dard in substation communications. Finally, several wireless technologies that have been tested on substation communications were presented as well. In the following chapter, a literature survey of the IEEE 802.11ax standard will be presented. What ensues in the third chapter is an attempt to test the suitability of IEEE 802.11ax for a smart grid substation scenario. That has been be con- ducted through simulations work. The final conclusions will be presented at the fourth chapter. Finally, a discussion regarding future related activity will be presented in the final chapter. The structure of the thesis report will be as follows:
• Chapter 2: A literature survey about the 802.11ax standard.
• Chapter 3: The simulation tool that was used and the simulation scenario that was implemented will be explained more thoroughly. Furthermore, simulation results for several scenarios will be presented.
• Chapter 4: Final conclusions regarding the network performance and the suitability of 802.11ax for smart grid substation communications will be given.
• Chapter 5: A discussion on future work, will be presented, including suggestions for other features of 802.11ax that have not been tested on this project.
Literature Study
2.1 The novel IEEE 80211ax standard
The needs for massive connectivity as well as higher data-rates have always worked as drivers for the development of new IEEE 802.11 standards. The first 802.11 amendment was released in 1997 and the offered data-rate was 2 Mbps. The next standards that were released were 802.11b with up to 11 Mbps, 802.11a/g with 54 Mbps, 802.11n with maximum data-rate of 600 Mbps, and finally, 802.11ac with above Gbps rates [13]. The increase of data-rates has been achieved with the adoption of higher modulation and coding schemes, as well as MIMO techniques. However, achieving even higher performance while using the legacy spectrum would require novel channel access methods, rather than wider spectrum band or more spatial streams. Furthermore, a high nom- inal data rate and sheer capacity do not necessarily represent the performance of the network, which could be affected by interference, frequency-selective attenuation and medium access issues.
In order to achieve higher performance, as well as deal with the aforemen- tioned inefficiencies, the IEEE Standards Association (IEEE-SA) approved IEEE 802.11ax in 2020. This standard aims to introduce PHY and MAC layer modifications that would enable at least four times improved throughput in a dense deployment scenario [14]. Multi-user access and energy efficiency need also to be achieved. These requirements are expected to be covered by the IEEE 802.11ax High Efficiency WLAN (HEW) technology.
According to [15], some of the key enhancements of this new technology are related to:
17
• Spatial reuse: dynamic adaptation of the PHY layer aspects, as well as the review of the beamforming implementation, so as to achieve more directive radiation patterns than the omnidirectional pattern used today.
• Temporal efficiency: the size of the header frame can be reduced with the utilization of channel reservation mechanisms (RequestToSend (RTS) and ClearToSend (CTS)). Another goal is to double the channel capacity through simultaneous transmission and reception of packets in different channels.
• Efficient channel monitoring: the carrier detection threshold of each Sta- tion (STA) can be adjusted with the use of Dynamic Sensitivity Control (DSC) algorithm that has been proposed, so as to avoid signaling pack- ages in a centralized way. This means that the detection range can be adapted dynamically, and thus deal with the hidden STA problem and improve the probability of successful transmission.
• Spectrum sharing: refers to the Dynamic Bandwidth Channel Access (DBCA) scheme, which was introduced in 802.11ac. Additionally, the Orthogonal Frequency Division Multiple Access (OFDMA) technology aims to improve spectral efficiency, as it divides the channel in several non-consecutive narrow-band subchannels.
• Multi-antenna technologies: the utilization of antenna arrays at both the Access Point (AP) and STAs leads to higher channel throughput.
The most important technologies introduced in 802.11ax will be presented in the following subsections.
2.1.1 PHY Enhancements
The modulation scheme is based on Orthogonal Frequency-Division Multi- plexing (OFDM), operating with 20 MHz, 40 MHz, 80 MHz, 80+80 MHz and 160 MHz channels. The OFDM symbol duration for the PHY payload has been quadrupled up to 12.8 µs, to render the signal more resilient to inter- user interference. That is highly important for UL MU transmissions which are executed by multiple users. Additionally, longer symbols lead to less over- head due to Guard Interval (GI). The available GI options are 0.8 µs, 1.6 µs and 3.2 µs, and the selection is conducted according to the channel conditions.
As a result, the overhead reduces to 6% from the 12-25% that was achieved in
the previous 802.11ac standard [13].
Besides the legacy BPSK, 16-QAM, 64-QAM and 256-QAM modulation schemes, 802.11ax introduces 1024-QAM which is meant mostly for indoors, high SNR environments. If combined with forward error correction codes, which have rates of 1/2, 2/3, 3/4 and 5/6, the data rates can reach a maximum of 9.6 Gbps [13].
Regarding the frame format, each frame begins with the preamble, which is duplicated on every 20 MHz channel [16]. As it can be viewed in figure 2.1, the preamble is comprised of two parts, the legacy and the High Efficiency (HE) part. The legacy part of the preamble is used for backwards compat- ibility with legacy STAs, while it also contains training sequences that are required for the synchronization of the receiver to the received signal. Finally, the legacy part contains the LSIG field that is used to calculate the frame dura- tion. The HE part of the preamble is meant to be decoded only by the 802.11ax stations. In order to be distinguished, it starts with the repetition of the LSIG field. Furthermore, the HE preamble comprises of the HE-SIG-A field, the optional HE-SIG-B field, as well as training fields needed for synchronizing MIMO.
Figure 2.1: The two parts of the preamble, the legacy and HE part, that are duplicated every 20 MHz [16].
More specifically, HE-SIG-A is two OFDM symbols long and it is dupli- cated in every 20 MHz channel. This field contains information meant for the reception and processing of the packet, such as the Modulation and Cod- ing Scheme (MCS) that was used, the bandwidth and the Number of Spatial Streams (NSTS) [16]. Additionally, HE-SIG-A contains information regard- ing the Basic service set (BSS) colour, the Transmission Opportunity (TXOP) duration, as well as the direction of the transmission (UL or DL). To ensure
that it will be received, HE-SIG-A may be repeated after going through an ad- ditional bit interleaving process, as it can be seen in figure 2.2.
Figure 2.2: Repetition mode for HE-SIG-A [16].
However, HE-SIG-A carries control information only for the case of Single- user (SU) transmission. Hence, HE-SIG-B has been introduced for the case of MU transmission, which contains both information for all STAs, as well as specific information for each STA.
Another new feature is the optional Dual Carrier Modulation (DCM). This mechanism sends the same signal on a pair of tones, that are not close in the frequency domain. This technology helps dealing with sub-band interference, while also introducing a 2 dB gain in the Packet Error Rate (PER) performance.
It reduces, however, the data rate twice as the data is duplicated [13].
2.1.2 Channel Access and Multi-user (MU) Transmis- sion
The Downlink (DL) Multi-user (MU) transmission feature, which was already added in the 802.11ac standard, is responsible for a significant increase in throughput, as it enables the AP to transmit data to various STAs simulta- neously. That is very helpful in scenarios where multiple spatial streams are difficult to establish. The Uplink Multi-user (UL MU) transmission is a newly added feature of 802.11ax [16].
Earlier versions of WiFi (802.11a and 802.11g), started to adopt OFDM since it is efficient for broadband communications. However, the channel was affected by frequency selective interference. To address this issue, the Orthog- onal Frequency Division Multiple Access (OFDMA) technology has been in- troduced. In this case, the sender can utilize only the best set of tones, which
consist a Resource Unit (RU), while the remaining tones are available to be used for other STA transmissions, hence introducing UL MU capabilities [16].
OFDMA transmissions are scheduled on a per-frame basis. This means that a frame can carry data to and from multiple STAs. In this case, various tones are assigned to several STAs, while maintaining the duration of each RU the same. Each RU may have 26, 52, 106, 242, 484, 996 or 2x996 tones (includ- ing service ones). The 20 MHz band, 40 MHz band, 80 MHz band and 80+80 (160) MHz band result in a 242-tone RU, 484-tone RUs, 996-tone RUs and two 996-tone RUs, respectively [13].
It is of high importance that MU-MIMO and OFDMA can be combined, which further enhances UL MU capabilities. That means that an RU can be assigned to up to 8 users. Additionally, each user can be assigned up to four spatial streams, given that the total amount of spatial streams is not greater than eight. For the case of DL OFDMA, the AP sends a MU packet with RU- dependent payload and with a common preample field HE-SIG-B that contains an RU allocation map, as well as indications as to how the RUs are assigned to an STA. The preample also contains the transmission parameters to be used by each STA (NSTS, MCS, coding scheme) and the spatial configuration to be used [13].
The UL MU transmission is a more tedious task. The AP transmits a Trig- ger frame, which is a new type of control frame, so as to specify several trans- mission parameters (duration, GI which is to be used by all the STAs partici- pating in the UL MU transmission) and ensure synchronization. Additionally, the trigger frame allocates RUs to each STA, and it also specifies several other parameters for each STA (such as MCS and coding). The UL MU transmission takes place after a SIFS, as it is depicted in figure 2.3 . Furthermore, the power of the received signal that is sent to the AP by the STAs needs to have almost the same power level. That is also achieved with the help of the Trigger frame that is sent from the AP, where the AP specifies its current transmit power, as well as the signal strength that it expects to receive from each STA. Therefore, the STAs can calculate the Path Loss to the AP, using the transmission power and the signal strength of the received Trigger frame [13].
OFDMA offers a significant performance improvement for the case of short packet transmission, as it reduces overhead caused by backoffs, pream- bles, interframe spaces and PHY headers, which carry common information for all the STAs during a DL transmission [16]. Additionally, the MU trans-
Figure 2.3: An UL OFDMA transmission [13].
mission needs to be aligned in the time domain. Hence, in the case that an STA has a very short packet to transmit, the STA may aggregate it with a new one. The 802.11ax enables to fragment the frames adaptively, so as to utilize the available space, as well as aggregate frames from various Access Categories (AC) during a MU transmission [16]. Furthermore, the hidden ter- minal problem is addressed with the use of MU-RequestToSend/ClearToSend (MU-RTS/CTS) handshake, which was introduced in the 802.11ac standard.
Furthermore, a new Multi-STA BA frame has been introduced, so as to deal with the overhead caused by acknowledging UL MU transmissions. It is used to acknowledge frames from several ACs. Finally, a MU Block ACK Request (BAR) has been defined, which can be used to request acknowledgements from several STAs in UL MU transmission.
Besides the scheduled OFDMA/MU-MIMO channel access in UL that has been mentioned, 802.11ax introduces an Aloha-like random OFDMA UL MU channel access mechanism [16]. It aims to cope with the case when the AP has no information regarding the UL traffic buffered at the STAs. In this case,
the AP sends a random access Trigger frame which allocates RUs for random access. Then, the STA that needs to transmit follows the OFDMA Back-off (OBO) process. This process determines if the STA is allowed to transmit, as well as the RU that is going to be assigned to the STA. During this procedure, the STA chooses a random value from the OFDMA contention window, which is defined in the Trigger frames sent by the AP. After every Trigger frame, the STA decrements the number of RUs specified in the Trigger frame from the initial random value, until it becomes zero. Afterwards, the STA selects a random RU and transmits its data. The Trigger frame may also improve the power efficiency of the system, by declaring when the next Trigger frame will be sent, and thus allowing the STAs to sleep most of the time, and wake up just before the Trigger frame for random access is transmitted.
2.1.3 Spatial Reuse and Overlapped BSS (OBSS) man- agement
In order to achieve higher performance in dense deployments, 802.11ax im- proves the Overlapped BSS (OBSS) function and spatial reuse. In order to decide whether the channel is busy or not, the STA senses the medium. The medium is considered busy when the STA receives a frame preample or in the case that it receives an unknown signal with power greater than 20 dBm above the minimum sensitivity threshold, or if there is indication that the channel is virtually busy [13].
The virtual carrier sensing feature in WiFi is called NAV. The NAV value is included in the MAC header originating from an STA, and it specifies how long the frame exchange will render the channel busy [13]. The rest of the STAs that receive this frame, consider the channel to be busy for as long as the NAV indicates. If an STA receives a frame with NAV value greater than its current NAV value, then the STA increases its NAV. The STA cannot decrease its NAV value, even if it receives a NAV value that is smaller. The STA can cancel its NAV only if it receives a CF-End frame.
In legacy WiFi systems, the STAs do not consider which BSS occupies the channel. However, a newly added feature of 802.11ax is that STAs can distin- guish if the channel is busy due to a transmission from an STA in its own BSS or an alien BSS. Then the STA can adjust its transmission power and the sen- sitivity threshold, as well as change its NAV value accordingly, so that more
efficient spatial reuse can be achieved.
What is more, the PHY preample contains information about the BSS colour, which is a non-unique identifier of the BSS. This feature was intro- duced in 802.11ah as a 3-bit identifier, and in 802.11ax it is increased to 6 bits, so as to reduce the probability of error, in case the colours IDs of two neighboring BSS are colliding. BSS colouring can help reduce the power con- sumption, since the receiver does not need to decode a frame that originates from an alien BSS [16].
Simulation Work
3.1 The Network Simulator NS-3
The Network Simulator 3 (NS-3) is a discrete-event network simulator for net- worked systems, that is widely used in research and for educational purposes in general. Its first stable version was released in 2008, with the aim to re- place the older Network Simulator 2 (NS-2). NS-3 offers a solid simulation core that is well documented and easy to debug, while also providing ease of configuration and trace collection. Furthermore, it enables the implementa- tion of models that are sufficiently realistic, so that NS-3 could be used as a real-time network emulator, that could be interconnected to the real world [17].
It can be used to simulate both IP and non-IP based networks, however it is mostly used for wireless/IP simulations. Such supported technologies are Wi-Fi, WiMAX, or LTE or other wireless systems for layers 1 and 2. There are further popular modules that include TCP performance monitoring and mobile ad hoc routing protocol modelling.
On the contrary to other network simulators that require a domain-specific modelling language to develop a network, NS-3 simulations can be imple- mented using C++ or Python, allowing users to benefit from the full support of each language. Finally, the object-modules that compose the network in NS-3, such as sockets and net devices, are aligned with a Linux computer, which enhances the realism and offers a better understanding of the system to the user.
According to [18], some key abstractions that are used for modeling in
25
NS-3 are the following:
• Node: it is the basic computing device. It could be considered as a com- puter, on top of which functionalities are added. These functionalities could include applications, protocol stacks or peripheral cards.
• Application: comprises an abstraction of the application level activity that is generated from the user program. Example of these applications could be client/server applications that generate and echo network traffic packets.
• Channel: models the media through which a device is connected to a network, such as a cable or wireless medium.
• Net Device: models both the hardware infrastructure (e.g. Network In- terface Card) and the software driver that would be required for a real device to be able to connect to a network. A Net device is installed on a Node, so as to allow it to connect with other Nodes through the Channel.
3.2 The Simulation Scenario
3.2.1 Deployment Scenario
This thesis aims at simulating the performance of IEEE 802.11ax for the trans- mission of IEC 61850 Sampled Values (SV) in substations. Specifically, we considered a setup in which several Merging Units (client devices, imple- mented as WiFi STAs) located at the process level of the substation commu- nicate with one IED (server device, implemented as WiFi AP) at the bay level over a wireless IEEE 802.11ax network.
The considered environment for the simulations is as is depicted in figure 3.1. The AP is placed in the middle of circular plane with radius R2. In this plane, several STAs are randomly deployed. To model the wireless propaga- tion environment, we have considered Line of Sight (LoS) conditions close to the AP and NLoS conditions far from the AP. Specifically, LoS connectivity is achieved inside the inner radius R1, where R1 = R2/2. The propagation parameters for LoS and NLoS have been derived by the measurements in [19], conducted in a 400kV substation, as depicted in figure 3.2.
Figure 3.1: The simulated deployment scenario.
Based on the measurements in [19], for the LoS case the chosen Path Loss Exponent (PLE) is 1.6, while for the NLoS case the PLE is 2.5. A piecewise path loss propagation model has then been implemented as in equation 3.1:
L =
L0, d < d0.
L0+ 10n1log(dd
0), d0 ≤ d < R1. L0+ 10n1log(Rd1
0) + 10n2log(Rd
1), R1 ≤ d < R2.
(3.1)
where L0 = 40.052 dB, d0 = 1 meter, n1 = 1.6, n2 = 2.5. In addi- tion to the path loss, and additional shadowing term has been added, in the form of a Gaussian random variable Xσ with zero mean and standard devia- tion σ = 2.76 dB [19].
In the simulations, different number of clients have been considered, up to a maximum of 10 clients, since a situation with 10 Merging Units connected to one IED represents the worst-case scenario in practical applications.
Furthermore, two scenarios were investigated:
• Small Substation: R2 = 40 meters
• Large Substation: R2 = 200 meters
These values have also been derived by practice, since 200 meters is often the maximum radius for large substations.
Figure 3.2: The outdoor 400kV substation scenario [19].
3.2.2 Traffic Model and WiFi Settings
As specified in the IEC 61850 standard, the SVs are modeled as Ethernet pack- ets whose minimum payload size is 64 bytes, which has been taken as the
default payload size in the simulations. According to the standard, each Merg- ing Unit generates 80 samples per power cycle (one power cycle is 50 Hz1 = 0.02 sec), which corresponds to a packet interval of 0.25 ms.
Regarding the WiFi settings considered for the simulation, the used stan- dard was "IEEE 802.11ax", at 2.4 GHz. The network performance has been tested for two channel bandwidths, 40 and 80 MHz. The chosen Modulation and Coding Scheme (MCS) was set to 8. The choice of the MCS will be explained more thoroughly later. The chosen Guard Interval (GI) is 800 ns, which is the smallest available option, and was selected to reduce the transmis- sion overhead. Finally, MIMO has also been tested, with two spatial streams.
The NS-3 classes that were used to implement the substation scenario are
"PacketSocketClient" and "PacketSocketServer". As it was depicted in fig- ure 1.4, the SV messages are directly mapped to the Link layer of the TCP/IP stack, and do not have any Transport or Network layer. The "PacketSocket- Client" and "PacketSocketServer" classes allow to generate raw sockets that link the Application layer directly to the Link layer of the stack.
The propagation delay and shadowing factor were modelled as described in the previous section. Each simulation had a 10 seconds duration, while re- sults from 10 runs were averaged for each configuration.
Finally, the performance of the network was measured also by using aggre- gation. Packet aggregation is defined as the transmission of multiple packets as a single transmission unit, with the aim to reduce the overhead of each trans- mission and the probability of collision in the network.
In the simulated implementation, the aggregation is executed at the appli- cation layer of the client. The aggregation number n is defined as the number of packets that are aggregated in a specific configuration. The tested aggrega- tion values are n = 1, 2, ..., 10, where n = 1 means no aggregation is used.
Specifically, to implement aggregation, the client waits for a specific period
of time before transmitting a packet, and this period is calculated as AggregInterval = P acketInterval∗n. Accordingly, the aggregated payload size is AggregSize = P ayloadSize∗n. The values of P acketInterval and P ayloadSize are 0.25 ms and 64 bytes, respectively, as reported at the beginning of this section.
A summary of the simulation configuration can be viewed in table 3.1.
Simulation Parameters
Deployment Scenario 1 server, up to 10 clients Scenarios Small substation (40 m radius)
Large substation (200 m radius) Propagation channel parameters n1 = 1.6
n2 = 2.5 Xσ ∼ N (0, 2.76) SV traffic modelling Packet interval = 0.25 ms
Packet size = 64 bytes WiFi settings Standard: IEEE 802.11ax
Frequency: 2.4 GHz Transmission Power: 20 dBm Channel Bandwidth: 40 MHz / 80 MHz
MCS 8 GI = 800 ns MIMO / No MIMO Uplink transmission only Aggregation n = 1, 2, ..., 10 Table 3.1: Simulation Parameters.
3.2.3 MCS selection
The Modulation and Coding Scheme (MCS) Index is defined by the number of spatial streams, the modulation type and the coding rate. Each MCS value corresponds to an expected transmission data rate. For example, according to the 802.11ax MCS table [20], when using OFDM modulation with 1 spatial stream, GI of 800 ns and channel width of 40 MHz, the minimum MCS value is MCS 0, which corresponds to the expected transmission rate of 17.2 Mbps, while the maximum MCS value is MCS 11, which corresponds to a data rate of 286.8 Mbps.
It needs to be noted that higher MCS values corresponds to higher data rate performance. However, this comes with a trade-off with regard to the Packet Error Rate (PER). When using a higher data rate, the probability of error is also higher, which can be observed also in figure 3.3. In order to overcome this issue, higher SNR at the receiver is needed. In figure 3.3 it can be observed
that for higher MCS values, greater SNR is required in order to minimize the PER. The figure was generated using the "ofdm-he-validation.cc" NS-3 script and setting WiFi parameters as in Section 3.2.2
-5 0 5 10 15 20 25 30 35 40
SNR (dB) 0
0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1
PER
PER vs SNR for MCS values 0-11
MCS value 0 MCS value 1 MCS value 2 MCS value 3 MCS value 4 MCS value 5 MCS value 6 MCS value 7 MCS value 8 MCS value 9 MCS value 10 MCS value 11
Figure 3.3: PER vs SNR for several MCS values (64 bytes frame).
The MCS to use in the simulated scenarios has been selected as follows.
Firstly, a calculation of the achievable SNR at the maximum distance (e.g.
200m or 40m) needs to be conducted, considering the adopted propagation model. For example, at 200m and with a bandwidth of 40 MHz, the SNR is 31.4 dB (considering maximum transmission power of 20 dBm and only path loss without shadowing). The next step is to derive the PER for frame sizes of 64 to 640 bytes (no aggregation up to aggregation n=10), for the obtained SNR and all the MCS values. Finally, the chosen MCS is the greatest MCS value for which PER=0 for the calculated SNR. In the simulated cases that consider channel width of 40 and 80 MHz, maximum distances of 40m and 200m and frame sizes from 64 bytes to 640 bytes, the optimal MCS values is found to be MCS 8, for all the mentioned cases.
3.2.4 Key Performance Indicators
In order to measure the performance of the WiFi network, and better compre- hend its suitability regarding the time critical requirements that the smart grid applications impose, several Key Performance Indicators (KPIs) were calcu- lated. These are:
• The mean value of latency E[latency], in seconds
• The Probability of Error (PER) for the packets
• The probability that the latency is greater than 3 ms P [latency > 3 ms]
• The standard deviation σ of the latency, in seconds
• Maximum Consecutive Lost Packets. This KPI is calculated as the high- est number of consecutive lost packets, after averaging the results that occur for the same simulation configuration. In the aggregated case, the calculation considers that for each lost aggregated packet, n packets are lost, where n is the aggregation number.
3.3 Simulation Results
The network simulator was configured as described in the previous section, and the simulations were conducted for a different number of clients, 1 up to 10. Performance metrics are going to be presented for both the large and the small substation case. Finally, the performance results with the use of aggre- gation are also going to be compared to the results without aggregation.
3.3.1 Results for the Large Substation case
Initially, the mean value of latency, as well as the PER, for the large substation case are presented in figures 3.4 and 3.5. The average latency is lower than 3 ms is satisfied for up to 3 clients for both 40 MHz and 80 MHz bandwidth, whereas for more clients the latency is above 3 ms.
The PER is following a similar trend as the average latency, and is higher than 4% for more than 5 clients. It can be noticed that both average latency and PER do not vary significantly between 40 MHz and 80 MHz channels.
This suggests that using the lower channel bandwidth (40 MHz) is preferable
due to the lower occupation of the spectrum.
1 3 5 7 10
Number of clients 0
0.002 0.004 0.006 0.008 0.01 0.012 0.014
Mean Value of Latency (sec)
Large Substation case - No aggregation
Latency 40 MHz Latency 80 MHz 3 ms threshold
Figure 3.4: Mean value of latency vs number of clients for a large substation.
Additionally, the probability that the latency is greater than 3 ms has also been calculated, and is presented in figure 3.6 for both the 40 and 80 MHz cases. It can be observed that for up to 3 clients the probability that the packet latency exceeds the 3 ms limit is less than 10%. However, this performance is decreasing significantly for a greater number of clients.
The performance improves significantly when aggregation is used, as it can be seen in figures 3.7 and 3.8. Additionally, the probability that the la- tency is greater than 3 ms is also decreasing after introducing aggregation, as it is shown in figure 3.9. Similarly to average latency, the best performances are achieved for aggregation numbers between n=4 and n=6, when the proba- bility of exceeding 3 ms drops around 10% from the 30% of the case without aggregation.
