Next-Generation Power Substation Communication Networks: IEC 61850 Meets Programmable Networks

The electrical grid is undergoing a fundamental change with the introduction of smart grid technologies. In particular, power substations have been evolving toward more automated systems. Power substation communication networks evolved from infrastructures mostly formed by serial devices to Ethernet-based digital communications networks, accelerating with the introduction of the IEC 61850 set of standards. However, this evolution inherited the shortcomings of the traditional decentralized network management. This article explores the upcoming evolution of IEC 61850 to meet the groundbreaking programmable network technologies: software-defined networks (SDNs) and programmable data planes (PDPs). Here, we describe how recent proposals leverage SDNs to improve network management tasks such as topology discovery, multicast traffic management, and quality-of-service (QoS) provisioning, among others in IEC 61850-based systems. We also outline potential improvements to critical network management tasks in power substations using PDP features such as in-band network telemetry. Finally, we discuss different challenges in the management of the communication networks of smart power substations and addressing them with the implementation of programmable networks.


Introduction
Power substations (substations for short) are critical infra structure elements in the provision of electricity service. Their main responsibility is the conversion of high voltages, present in the generation and transmission process, to lower voltages connecting to end users. Substations integrate volt age levels using transformers that perform voltage stepup or reduction and switches used for system reconfiguration, including circuit breakers (CBs) for performing protec tion operations. Energy markets are evolv ing to meet specific customer requirements, for example, through payperuse services where end users can pay for the exact amount of electricity used. New markets facilitate the participation of different players, includ ing those contributing as energy generators by supplying excess energy from their own generation (such as small wind generators or solar panels) to the grid, supported by utilities introducing customer and supplier satisfac tion as an objective. This satisfaction relies on information exchange among the entities involved in the process of energy supply, which depends on a set of mechanisms that enable users to decide how to plan their energy demand. Hence, the devel opment of architectures to better facilitate that exchanged information is very relevant. The increasing size and com plexity of the communication infrastructures within power system substations makes management and operation based on human intervention unfeasible. The criticality of these infrastructures, and the challenges associated with resil ience, robustness, availability, and security, imply that the management of infrastructures needs to evolve toward auto mated models based on the exploitation of data available from entities connecting to the infrastructure.
To achieve the purpose of having such architectures while considering design principles such as vendor indepen dence and openness, Technical Committee 57 of the Inter national Electrotechnical Commission developed the IEC IMAGE LICENSED BY INGRAM PUBLISHING 61850 family of standards for the specification of communi cation protocols and data models for use in substations. An important purpose behind the introduction of this standard was to enable evolution from legacy infrastructures, mostly formed by serial devices interconnected through complex wire meshes, toward digital communication networks, based on wellknown technologies such as Ethernet, to enable flex ible service models tailored to the satisfaction of different customer needs.
Despite the benefits of using Ethernetbased commu nication, the technology introduces operational issues and security concerns. If not implemented well, the use of digital communication networks limits scalability of the infrastruc ture and requires manual configuration of each networking device, thus increasing the complexity of the communication network management. IEC 61850 relies on configurations such as IEEE 802.1Q virtual local area networks (VLANs), and it does not incorporate security mechanisms by design. Hence, the communication infrastructure of IEC 61850 pres ents limitations to its scalability by relying on VLANs as the mechanism for traffic segregation and is prone to security threats such as replay attacks, false data injections, spoofing, and denial of service, among others.
Two recent paradigms seeing application in enterprise computing applications have the potential to reshape substa tion computer networks. On the one hand, SDNs have intro duced decoupling of the control plane, where algorithms and logical functionalities of the network reside, from the data plane where the actual packet forwarding occurs. On the other hand, PDPs have enabled customization of the pars ing and processing of network packets within switches and routers through the deployment of chips that end users can program (rather than configure). Similar to the approach fol lowed in different computer network contexts, researchers are exploring SDNs and PDPs as mechanisms to cope with the security and operational issues that might arise in IEC 61850 infrastructures. For example, some proposals aim at exploiting the functionalities of SDNs to support the network segmentation required to create different broadcast domains to separate the multicast groups associated with different control functions.
In recent times, SDNs have seen implementation in smart substation communication networks. This incorporation might bring several advantages for network management in IEC 61850 infrastructures. For instance, leveraging SDNs provides isolation and slicing to separate the different traffic types in the communication traffic to provide improved QoS. SDNs can also prioritize the traffic associated with critical events in the substation and might simplify network manage ment due to the separation of control and data planes. More over, SDNs can provide mechanisms to ensure the resilience of substations by seamlessly rerouting or duplicating traffic upon the failure of the control devices.
This incorporation of SDNs in power systems is in active development by industry, with success cases that confirm its potential. In their 2020 paper, engineers from Schweitzer Engineering Laboratories (SEL) discuss the engineering process for incorporating SDNs in a complex, thermal based power generation infrastructure in India. The authors describe an infrastructure integrating more than 60 intelli gent electronic devices (IEDs) with two supervisory control and data acquisition servers and their respective humanmachine interface (HMI) systems. These authors intro duce the concept of operational technology SDNs using a proactive configuration of flow rules to define the network's behavior. According to their analysis, SDNs outperform traditional network approaches by reducing failover times regardless of the network topology.
In addition, with SDNs, it is possible to implement net work packet filtering to provide improved security, traffic forwarding based on a matchaction model, a security model based on a denybydefault approach, and proactive behav ior for dealing with failover events. In addition to this, SEL also reports two other important study cases based on SDNs. In one case, the U.S. Department of Energy's National Renewable Energy Laboratory chose a solution based on SEL technology for the Energy Systems Integra tion Facility, located in Colorado, which includes a research platform for distributed energy and microgrid devices and systems. The solution implemented by this vendor incor porates SDNs for its microgrid controller platform to improve security and manageability of the infrastructure. In another case, they deployed a solution based on SDNs for the operation of Itaipú Dam, operated between Bra zil and Paraguay. According to the white paper describ ing this solution, SDNs provide optimized traffic man agement, enhanced response to operational events, fast failure recovery (which is critical for such infrastructure), improved cybersecurity, and more precise testing and doc umentation of operation tasks.
Finally, researchers at the Research Group on Applied Telecommunications at the University of Antioquia in Colombia present a reference architecture for the manage ment of communication networks of substations. This archi tecture enables an improved approach to security, QoS, and fault tolerance by leveraging the previously mentioned properties associated with SDNs. A certification laboratory validated and assessed this architecture. The certification laboratory allows utility companies to verify the compliance of their network deployments with the IEC 61850 standard, showing important results regarding operational conditions of the infrastructure.
The concept of PDPs has emerged as a complete real ization of the original ideas introduced with SDNs. PDPs enable the network owners to specify details of the protocols managed within switches and the actual treatment applied to packets. These programmable forwarding devices can apply custom processing to provide specific functionalities that match the needs of traffic in smart substations. Despite the advantages and the possibility of developing custom packet processing mechanisms, to the best of our knowledge, PDPs have not yet been widely applied in the communication net works of smart substations.
In this article, we present a discussion of the potential benefits that the integration of programmable networks introduces to the communication networks of smart substa tions based on IEC 61850. The rest of this article presents the following topics. The "IEC 61850 Digital Power Sub stations" section presents the background on the architec ture and protocols integrating digital power substations. The "Network Programmability in IEC 61850" section discusses the possibilities of applying network program mability in the communication networks of smart substa tions. In the "Challenges and Future Perspectives" section, we state some challenges that need to be addressed when considering network programmability in smart substations and the future areas for research that we foresee to com plete this integration.

IEC 61850 Digital Power Substations
A power substation is an important component in the chain of generation and supply of electric power as it is in charge of the transformation and distribution of electric energy. A digital power substation is a power substation that fully incorporates the advantages of data networks to improve operation and maintenance in the areas of protection, control, communications, and monitoring. The operation of a digital power substation involves multiple Internet Protocol (IP)compliant devices that are interconnected through a multilevel network infrastructure based on Eth ernet technology with the purpose of forming a commu nication platform supporting management, monitoring, synchronization, protection, control, and sensing opera tions (see Figure 1).
A digital power substation is a modern form of a sub station automation system (SAS), where the concept of an automation system implies integration of the automated responses of the control, protection, and monitoring processes within this infrastructure as a holistic system. In modern context, the IEC 61850 framework can govern the substa tion automation process, covering almost all aspects of an SAS to guarantee interoperability of devices from different manufacturers. The standard also defines how management, control, protection, and measurement devices intercommu nicate inside a substation. As shown in Figure 2, the model proposed within IEC 61850 is hierarchical with three levels: station, bay, and process. These levels are interconnected via the process and station buses.
The process level is composed of yard equipment [voltage transformers (VTs), current transformers (CTs), and CBs], measurement processing devices [merging units (MUs)], actuators (IEDs) for controlling CBs, and time synchroniza tion (TS) devices (main clock). MUs are devices in charge of acquiring measurements of currents and voltages from the instrument transformers (VTs/CTs), converting these analog signals to digital, performing signal processing and subse quently transmitting them over an Ethernet network accord ing to guidelines outlined in the IEC 61850 standard.
The bay level, located near the yard devices, includes relays, referred to here as IEDs, for protection and control at the physical level. These IEDs continuously process infor mation generated from the MUs to provide continuous and detailed knowledge of the different events occurring in the substation or surrounding power system. The IEDs identify events and take appropriate actions (for example, trigger ing an actuator to automatically trip CBs for a power line due to the presence of a fault, generating alerts upon indi cations of problems with the health of equipment such as switches or transformers, or isolating faulty sections of the power network).
The station level provides an overview of the entire SAS, including an HMI workstation that presents the processed data for a local operator to monitor and interact with the different substation processes. At this level, there are also devices that provide connectivity between the power substa tion and the control center.
There are two physical subnetworks, formed by Ethernet switches, which interconnect the three levels. The station bus provides connectivity between the stationlevel and baylevel devices, typically implemented in a ring topology. The process bus interconnects the processlevel equipment with the bay level equipment by leveraging resiliency techniques such as the Parallel Redundancy Protocol (PRP), which is a simple proto col used for redundancy in industrial networks, based on the notion of physical redundancy. PRP implements redundancy in a network in the form of redundant switches linked with two independent physical paths. The switches send packets across the two paths simultaneously, and the receiving device accepts the first packet to arrive and discards the second one. If one of the paths fails, packets travel over the other path to guarantee failtolerant operation of the network. The IEC 61850 standard also defines four types of com munication services: an abstract communication service inter face, a generic objectoriented substation event (GOOSE), SVs (sampled values), and TS to ensure the correct operation of a fully digital substation (see Table 1). Substation owners may also choose to implement a subset of these services, for example, utilizing only GOOSE as an intermediate step toward a fully digital substation.
According to the IEC 618505 and IEC 618508 recom mendations, the communication services are mapped into different communication stacks according to their perfor mance requirements (see Figure 3). For example, the Manu facturing Message Specification (MMS) is transported over IP, whereas GOOSE and an SV are transported directly over Ethernet frames transmitted via multicast.

Network Programmability in IEC 61850
As mentioned earlier, IEC 61850 mainly concentrates on the digitization of substation communication networks by means of a transition to Ethernetbased systems. Therefore, the adoption of IEC 61850 has propelled the modernization of substation communication systems. Adopting such systems, however, inherits their historical management complexity.
The variety of communications protocols (e.g., SV, GOOSE, MMS, Precision Time Protocol, and DNP3, among others) further complicates network management. For instance, SV and GOOSE rely heavily on data link multicast (transmis sion to a given set of devices rather than to every device connected to the medium), forcing network devices to be On the other hand, network programmability has emerged to provide flexible and customized network management. This recent paradigm for network programmability uses two architectural proposals: SDNs and PDPs. SDNs separate con trol and data planes and provide a logically centralized control of the network through a programmable control plane. More recently, the emergence of PDPs represents an outstanding advance in the complete realization of the SDN paradigm. PDPs enable complete control of network behavior, from the applications to the packet processing within the forwarding devices, including the definition and parsing/deparsing of custom headers. Consequently, PDPs allow revisiting exist ing functions for network management. Figure 4 illustrates a comparison of the evolution of the architectures from legacy switches (incorporating control and data planes in the same device) toward a programmable environment using an SDN based architecture using OpenFlow and P4based switches. Figure 4 depicts several aspects illustrating the contrast between traditional networks and programmable networks. On the left side, we present the conceptual structure of traditional network devices. Traditional network devices have tightly cou pled and embedded control and data planes. The control plane contains the programs that execute the algorithms involved in tasks such as routing, QoS, and security. These algorithms operate in a distributed way, exchanging information with the corresponding instances running on other devices in the net work. The data plane, on the other hand, is only in charge of performing packet forwarding, which might include packet header updating and calculation of checksums. On the right side, we depict elements of a network paradigm based on the combination of SDNs and PDPs. First, SDNs enable the decoupling of control and data planes, which is not possible in conventional forwarding devices. SDNs separate these planes and migrate the control plane out toward a logically central ized entity called a controller or network operating system. Examples of controller implementations are NOX, POX, and Ryu (implemented in Python) as well as ONOS and Open DayLight (implemented in Java). Network functionalities such as routing become software applications developed using generalpurpose languages matching the implementation lan guage of the controller. The applications implementing those network functionalities run in the controller and communicate with the data plane through a set of communication interfaces. Then the forwarding devices perform the actual packet for warding. The most developed of the interfaces between the control and data planes is the OpenFlow Protocol, specified by the Open Networking Foundation, which has become the de facto standard in SDNs.
PDPs complement the concepts proposed by SDNs, pro viding a complete realization of the SDN paradigm. PDPs propose programming the packet parsing and processing within forwarding devices according to custom needs speci fied by network users through specialized programming languages. For instance, the P4 programming language, introduced by researchers at Princeton University in 2014, has become the de facto standard for data plane program mability. In data plane programmability, equipment vendors provide compilers that translate the userwritten P4 pro grams into machine code for the specific target chip in the switch (represented in Figure 4 by the executable and p4 info files, which are output by the compiler). This translation also provides an interface called P4Runtime, which the control

Type of Service Description
ACSI Defined in IEC 61850-7-2, the ACSI addresses the basic requirements for the process of information exchange. With this aim, the Manufacturing Message Specification transports operational information for the management of the substation between the user interface system in the control center and the IEDs, including the monitoring processes.

GOOSE
Defined in IEC 61850-8-1 for the purpose of distributing event data (commands, alarms, status information, and trip messages) among IEDs across the entire substation network.

SV
Specified in IEC 61850-9-2, the SV transmits analog values (current and voltage) from the MU to the IED. plane can use to control the actual packetforwarding task provided by the switch. Until recently, the focus of SDNs and PDPs was on enterprise network environments such as Inter net service providers, wide area networks, wireless 5G and beyond, and, especially, data center networks. Leveraging SDNs and PDPs can achieve this automation of the different aspects of the network management in the context of digital power substations based on IEC 61850. To cope with the aforementioned management issues, recent installations have included SDNs as a central element of the IEC 61850 network architecture. In fact, manufactur ers of power systems equipment, such as SEL, have deployed (proprietary) SDNbased solutions for power substations' communication networks. Figure 5 shows an SDNenabled architecture where a net work controller acts as a programmable control plane that enables automated communication among the IEC 61850 architectural levels. Several proofofconcept approaches have proposed different contributions for automating IEC 61850 communication network management with SDNs. The following are important network management functions automated by SDNs:

TS
✔ Network topology discovery: OpenFlowenabled switch es are configured to read Link Layer Discovery Proto col messages to provide a global view of the network topology to the network controller. ✔ Multicast traffic management: This function handles multicast traffic (GOOSE, SV); automatic slices (i.e., logically isolated network partitions) are set up by cal culating optimal multicast routing trees in the network controller. ✔ Prevention of switching loops: Often, implementing the Spanning Tree Protocol (STP) prevents dataflow loops that hurt network performance. However, use of the STP limits network scalability as it allows only one active path between two nodes. In contrast, SDNs allow the use of optimization techniques to find the best group of loopfree paths as the SDN controller is fully aware of network topology. ✔ Redundancy: To guarantee zero recovery time in criti cal services (such as GOOSE and SV) upon node or link failures, the PRP or the High Availability Seam less Redundancy (HSR) Protocol are used in the in dustrial Ethernet, including IEC 61850. However, the PRP duplicates the original LAN of the power sub station, increasing the capital expenses (CAPEX); and the HSR needs ring network topologies and spe cial endnode devices. In contrast, an SDNenabled strategy enables building a zero failover approach that builds two link and nodedisjoint trees for each GOOSE or SV service in the network controller. ✔ Provision of QoS: QoS is commonly implemented us ing class of service values to assign priorities to the VLANs carrying the traffic, usually resulting in over provisioned networks. In contrast, SDNs can enforce QoS policies by identifying the critical flows and ap plying queuing disciplines to guarantee that their ex pected constraints are met (e.g., a maximum of a 3ms delay for GOOSE trip messages). PDPs are a natural step beyond OpenFlow in SDNs. PDPs provide network reconfiguration capabilities (the control ler can redefine packet parsing and processing in the field), protocol independence, and target independence. These capabilities enable novel features such as inband network telemetry, an approach where the network packets can contain statistics updated upon the processing of these pack ets within the network devices, allowing customized net work management. In substation communication networks, the introduction of PDPs would help with many network management challenges, such as cybersecurity, congestion  control, improving QoS, network infrastructure awareness, and management automation, among others. The next sec tion describes in detail these important challenges and dis cusses potential alternatives to face these challenges using programmable networks.

Challenges and Future Perspectives
The incorporation of PDPs in the communication infra structure of smart substations introduces several chal lenges, especially in the context of network management. Next, we discuss some of these challenges and then out line the possible approaches to address them via program mable networks.

Cybersecurity
The incorporation of SDNs and PDPs in the network of smart substations introduces several possible ways to imple ment security measurements for these infrastructures. On the one hand, the logical centralization and global visibility of the network provided by SDNs allows effective deploy ment of applications and algorithms that make security decisions, such as traffic blocking, network segregation, or rerouting. On the other hand, programmable switches have detailed visibility of traffic and enable the ability to perform quick actions on the traffic due to their location. However, despite this advantage, the need for coherence and prompt response without inducing excessive overhead in the traffic processing is a challenge for implementing these security actions. A potential approach to enable intru sion detection inside data plane programmable switches makes use of lightweight machine learning techniques (i.e., binarized neural networks) and shows great improvements in reducing latency and communication overhead over edge network domains.

Congestion Control
One of the main applications of PDPs, as reported in the lit erature, is inband network telemetry. This application takes measurements of the packets within switches and passes the information derived from these measurements up to the con trol plane. Thus, the control plane can make decisions by ana lyzing this information in light of a global view of the network topology. By incorporating these measurements in this analy sis at the control plane, aspects such as network congestion can be managed. For example, congestion can by decreased by creating alternate paths that guarantee that critical messages (e.g., type 1 A GOOSE messages) do not become affected by increments of delay due to congestion events.

QoS
Communication among devices in smart substations uses pro tocols such as GOOSE and SV. These protocols in general, and GOOSE in particular, define different types of messages according to the information related to the events that might occur in the infrastructure. For example, there are classes of GOOSE messages associated with critical operations in the infrastructure that have very strict requirements regarding delay. Hence, it is desirable to have the capability to prioritize and provide a differential treatment that can privilege to these critical messages over information or monitoring traffic. The combination of SDNs and PDPs can contribute to address ing this challenge by seamlessly configuring dedicated paths through devices capable of distinguishing and performing expedited forwarding of critical packets.

Infrastructure Awareness
An important aspect of the communication infrastructure in smart substations is communication resilience in terms of providing alternate communications paths upon the failure of critical nodes. Introducing redundant paths by duplicat ing infrastructure improves resilience but also increases CAPEX and operational expenses and increases network complexity. By leveraging SDNs and PDPs, it is possible to avoid the need to duplicate infrastructure. Networkresil ience improvement then results from taking advantage of the global visibility of the network topology, which is inherent to SDNs. This global visibility, in combination with function alities such as inband network telemetry, might be useful to detect the degradation of devices connected to the network in advance of failure, while also providing, in advance, alter nate paths for communication among critical devices. This guarantees continuity of the network operation by carefully observing the network topology behavior with SDNs and observing the particular dynamics of the network traffic by leveraging PDPs.

Network Management Automation
In the pursuit of true resilient and trustable infrastructures, which are vital in the context of critical infrastructures, automation of network management operations is a vital need. For instance, the collection of statistics to support decision making, possible forecasting of failures, or ser vice degradation is a fundamental task. As discussed previ ously, programmable devices can implement inband network telemetry. Thus, network management systems can access information that is more accurate rather than gathering sta tistics through periodic polling. In addition, the provision of "firsthand" measurements from network devices can be leveraged to design resilient and proactive security mecha nisms based on data analytics implemented at the control plane. By using these data analytics, it is possible to antici pate breakdowns and react to them in advance by enabling alternate communication paths to overcome critical situa tions such as attacks or communication failures.
Substation management automation is paramount in the migration toward a nextgeneration network core (i.e., net work infrastructure leveraging SDNs, PDPs, and manage ment based on data analytics to support automation for the communication of core operations in the substation). In this environment, the new architecture needs to support the dynamic implementation of a variety of different function alities, such as cybersecurity. The application of automatic software vulnerability management and security patch updates for substation products are essential security measures in the communication environment. In the same way, critical tasks such as updating the firmware of different families of switches and the reception and execution of actions in response to alerts from the manufacturers must occur without human intervention.
The market needs an open ecosystem for substation con trollers and engineering tools, especially when proprietary solutions are the common scenario today. Another issue of great interest is automating policy management for threat intelligence to increase the transparency of monitoring tasks. Additionally, information sharing between network func tionalities such as intrusion detection systems or security information and event management needs continuous updat ing with threat intelligence information.

Fulfilling of Critical Time Requirements
Within the hierarchy of IEC 61850 messages, there are dif ferent priorities according to their use cases. In particular, the trip messages (type 1 A GOOSE), associated with com mand or status notifications and the raw messages (SV, type 4) require transmission times between 3 and 10 ms. These strict time requirements imply that any processing performed on the packets associated with these messages must not induce overhead that could compromise these time requirements. Despite the advantages PDPs introduce in terms of flexibility and expressiveness through custom packet processing, there exists a critical tradeoff between these advantages and the stringent time requirements defined by parts of IEC 61850.

Particularities of Infrastructure
In general terms, there is no concept of a "standard substa tion." Actual deployments might differ in the topology, con figuration, number of instances of devices (IEDs, MUs, and actuators), and vendors of these devices. Despite the fact that IEC 61850 is an initiative for standardization and interoper ability, some aspects such as the actual capabilities avail able in devices might vary. Hence, some particular aspects of the network communication, such as the fields contained in GOOSE/SV frames and their interpretation might be dif ferent across different deployments. This fact constitutes a challenge because then, each deployment and the implemen tation solution based on network programmability needs to be customized according to the particular elements applied for such a solution.

Lack of Datasets
One of the areas that might become important to take advantage of in network programmability in smart substations is secu rity. The research literature presents a wide set of proposals to develop intrusion detection and prevention systems for substa tion infrastructures. Leveraging features provided by SDNs and PDPs can improve applying these security solutions. However, an important challenge in developing potential machine learning solutions is having datasets to train the models for these solu tions. According to the literature, most of the approaches use either private datasets or generalpurpose datasets not tailored to the protocols of smart substations. There are only a couple of public datasets, such as EPIC and RICsel21. However, not all information in these datasets is available, especially the informa tion associated with the particular capabilities of the devices used to obtain the data. As a result, the development of general solu tions for security of smart substations based on machine learning and leveraging programmable networks might need to rely on transfer learning (an artificial intelligence technique that uses a model previously trained in a different domain and applies it to a different one) to achieve a certain level of generality.

Need for a Standardized Risk Management Framework
In addition to taking advantage of the potential for integrat ing SDNs and PDPs in the core of substations, it is essential to develop a framework that simultaneously accomplishes the security requirements of IEC 61850 and those coming from the Internet world (i.e., a framework capable of addressing the tradeoff between meeting the strict requirements of IEC 61850, especially in terms of processing times for packets, along with the computation time required by machine learn ing techniques by leveraging programmable networks). Risk management in smart substations might leverage the proper ties of traffic visibility and logical centralization provided by SDNs and PDPs to improve security and resilience, but it must assure that enabling this property does not hamper the correct processing of critical GOOSE or SV messages.

Concluding Remarks
In this article, we outlined some elements in the evolution of the communication infrastructure of smart substations. We dis cussed some of the basic elements of IEC 61850, concentrat ing on its network architecture and operational requirements, and the resulting communication needs in substations. We also presented some operational challenges that have been brought by digitization of the communication infrastructure of smart substations. We consider that the combination of SDNs with PDPs provides a big opportunity to introduce novel and effec tive solutions to address these operational challenges. How ever, two aspects of applying these tools are still open research problems. The first is the actual details of implementing these solutions. Determining which logic to implement at the control plane versus what to offload toward PDPs is not a trivial prob lem. A second challenge is the tradeoff between granularity and visibility of traffic processing based on functionalities at the PDP versus the strict delay requirements defined for traffic in power substations. Hence, in the short term, there will be an increasing interest from academia and industry in further inves tigating these open research problems.