Tailoring DDS to Smart Grids for Improved Communication and Control

Alaa Alaerjan and Dae-Kyoo Kim

Department of Computer Science and Engineering, Oakland University, Rochester, Michigan, U.S.A.

Keywords:

DDS, Smart Grids, Communication, System Reliability, QoS.

Abstract:

Adopting new technologies in smart grid (SG) enables the improvement of reliable communication. A key fac-

tor for SG efﬁciency is reliable data exchange between different components and domains in the system. SG

must allow remote and quick reaction for different events. This is not a trivial task especially with large scale

power grids, which requires SG to have a reliable communication protocol. Data Distribution Service (DDS)

is introduced as a data-centric middleware standard based on publish-subscribe protocol to address communi-

cation needs for distributed applications. DDS supports reliable data exchange between different components

using various features such as quality-of-service (QoS). In this paper, we describe how DDS can be tailored to

support SG to improve the communication of devices in SG. We ﬁrst give an overview of DDS and discuss the

beneﬁts of applying it to the communication system in SG. We then describe communication requirements and

constraints in SG. and discuss how DDS can be tailored to SG with respect to the requirements and constraints.

1 INTRODUCTION

The traditional power grid uses the simple power gen-

eration and consumption paradigm which involves lit-

tle management for efﬁciency, and thus has signiﬁcant

power loss. According to the report by The World

Bank (World-Bank, 2015), the U.S loses 6% from

its total power in transmission and distribution pro-

cess, the U.K loses 8%, China loses 12%, some other

countries even lose a signiﬁcant amount, such as Iraq

which loses about 35% of its produced power. Fur-

thermore, traditional power generation systems create

a large amount of carbon dioxide (CO2) contaminat-

ing the environment.

Smart grid (SG) has emerged as the next gen-

eration for improved efﬁciency of power production

and consumption. Unlike the traditional power grid

which is not designed for device communication, SG

aims at facilitating communication and data exchange

between various equipment and devices across the

power domain. It enables bidirectional data exchange

between powersuppliers and consumers for improved

power management (Fang et al., 2012). As an exam-

ple, in SG, smart meters measuring power consump-

tion are able to communicate their data to utilities

which are in turn able to send real-time pricing back

to smart meters.

Communication management in SG is a challenge

due to the heterogeneity of the infrastructure. SG

components involve signiﬁcant data exchanges re-

quiring efﬁcient communication. A given approach to

enhance SG communication has to consider concerns

such as time constraints and devices requirements.

Some protocols such as the Distributed Network Pro-

tocol (DNP3) (NIST, 2010) and Modbus (Modbus-

IDA, 2006) have been tried to overcome the commu-

nication challenges in SG. However, they introduce

signiﬁcant overheads and latency which make them

unsuitable for SG (discussed more in Section 2).

DDS has emerged as a potential model to ad-

dress the challenges in SG. Based on a simple pub-

lish/subscribe protocol and QoS policies, it is de-

signed to support high-performance, scalable, de-

pendable and real-time data exchange between dif-

ferent components with little overheads. In this pa-

per, we describe how DDS can be tailored to SG. We

ﬁrst identify communication requirements for differ-

ent components in SG and discuss how DDS should

be tailored to satisfy those requirements.

The remainder of the paper is organized as fol-

lows. Section 2 gives an overview of the DDS model

and describes the advantages of applying it to SG.

Section 3 outlines the related research and describes

how the work in this paper is different. In section 4,

we describe the SG communication requirements for

different domains in the system. In section 5, we de-

scribe how to tailor DDS to SG to achieve improved

communication and control. Finally, Section 6 con-

cludes the paper and outlines the future research.

Alaerjan, A. and Kim, D-K.

Tailoring DDS to Smart Grids for Improved Communication and Control.

In Proceedings of the 5th International Conference on Smart Cities and Green ICT Systems (SMARTGREENS 2016), pages 433-438

ISBN: 978-989-758-184-7

433

2 OVERVIEW OF DDS

Data Distribution Service (DDS) (Object-

Managment-Group, 2015) is a data-centric mid-

dleware standard by the Object Management Group

(OMG). It came in 2004 as a Publish/Subscribe

protocol to address the data sharing needs for a

wide variety of computing environments, ranging

from small local networks to large scale systems.

DDS provides a scalable platform and location-

independent infrastructure to connect publishers and

subscribers. DDS also supports a wide variety of

quality-of-service (QoS) properties including time

sensitivity, reliability, and many others (Corsaro,

2014).

The DDS speciﬁcation is described as two-level

interfaces. The ﬁrst level is Data-Centric Publish-

Subscribe (DCPS) which is a lower layer API that al-

lows DDS-enabled applications to communicate with

each other. The second interface is Data Local Re-

construction Layer (DLRL) which is an upper layer

speciﬁcation that outlines how an application can in-

terface with DCPS. DLRL is deﬁned as optional.

DCPS is the core of the DDS model which allows

heterogeneous components to interact through read-

ing and writing the data form/to the global data space.

Components interact by declaring their intent to pub-

lish or subscribe data. Figure 1 shows the interaction

(through a global data space) between different enti-

ties in the model. DDS entities that are involved in

DCPS-based applications include:

• Domain: This is a conceptual container in the

system. Components in a domain can only com-

municate within the same domain, which allows

for data isolation and optimized communication.

• Domain Participant: This represents the partici-

pation of a given application in the domain. Each

process has one domain participant for each data

domain.

• Data Writers and Publishers: An application

uses a writer to write the data. It is also an in-

Figure 1: DDS architecture and components.

terface (access point) to publishers. A publisher

is an entity that is responsible for data issuance to

the data domain. A publisher can be used by an

application to group multiple writers.

• Data Reader and Subscriber: A reader is uti-

lized by an application to receive the data. It is

also the access point to subscribers. Similar to

publishers, a subscriber can be used as a container

to group multiple readers.

• Topic: This is a basic data object to be pub-

lished/received. A given topic must match in or-

der to establish a connection between a publisher

and a subscriber. A topic is deﬁned in terms of

name, data type, and QoS.

DDS has many advantages over other pub-

lish/subscribe protocols such as DNP3 and Modbus.

First, DDS is simple and ﬂexible, while supporting

complex messaging scenarios as opposed to Modbus.

Second, unlike DNP3 which induces 50%-80% of

the processing delay in embedded power devices (Lu

et al., 2011), DDS does not introduce any overheads.

DDS also enables high-performance data exchange

and QoS on a component base, while supports differ-

ent types of communication such as one-to-one, one-

to-many, many-to-many, and many-to-one (Object-

Managment-Group,2015). These features make DDS

as a potential solution for addressing the communica-

tion challenges in future grid.

3 RELATED WORK

Smart grid is still in its infancy stage. Organizations

and researchers have been trying to use and integrate

different protocols to improve the communication

system in SG. Some works in the literature (Bakken

et al., 2009; A. Alkhawaja, ; Twin Oaks Comput-

ing, 2011) have considered use of middleware solu-

tions such as DDS or message-oriented middleware

for communication and data exchange in SG.

The work by Bakken et al. (Bakken et al., 2009)

argues that middleware solutions (e.g., DDS frame-

work) are better approaches to address interoperabil-

ity and data exchange in SG. They justify the advan-

tages of middleware standards over other protocols on

achieving reliable end-to-end communication. Carry-

ing over their study, in this work, we address what

needs to be done in order to adopt DDS to SG.

Alkhawaja and Ferreira (A. Alkhawaja, ) also con-

sidered the integration of DDS with the SG domain.

Their work discusses use of existing middleware so-

lutions to support distributed large-scale applications

with QoS requirements. In the same light as Bakken

et al.’s work, their work also emphasizes the bene-

SMARTGREENS 2016 - 5th International Conference on Smart Cities and Green ICT Systems

434

Figure 2: Smart Grid Communication.

ﬁts of applying DDS to the SG. They show that the

lightweight architecture of DDS ensures high perfor-

mance and predictability by its capabilities to reserve

resources and enforcing QoS. However, the scope of

their work does not considers SG requirements and

tailoring DDS to SG. In our work, we address how

DDS should be tailored to satisfy SG requirements.

The report by Twin Oak Computing (Twin

Oaks Computing, 2011) describes the capabilities of

DDS in general. It describes how the DDS architec-

ture can improve communication in large scale sys-

tems such as power grid systems. The report argues

that DDS can be used for safety critical systems such

as renewable energy systems. It also describes the

importance of DDS in achieving communication in-

teroperability. However, there is little discussion as to

use of DDS in SG.

4 SG COMMUNICATION

REQUIREMENTS

Smart grid involves four main domains (NIST, 2010)

– i) power generation, producing power on high volt-

age levels, ii) power transmission, transmitting gen-

erated power to substations, iii) power distribution,

distributing power to end users, and iv) power con-

sumption, ultimately consuming the power. Each one

of these domains has its own components and require-

ments. Figure 2 depicts the communication involved

in the domains (NIST, 2010). In this paper, we use the

term component interchangeably with device.

Before adopting DDS to SG, the requirements of

SG communication should be identiﬁed, which pro-

vides a base for implementing DDS. As discussed in

Section 3, there are some works that discusses pos-

sible implementation of DDS in SG. However, to the

best of our knowledge, there is no work addressing

communication requirements and how DDS should

be tailored to SG to satisfy the requirements. In this

work, we study different types of devices involved

in SG communication in the above four domains and

identify communication requirements and constraints

imposed by the devices. In the rest of this section,

we discuss the four domains in terms of involved de-

vices and related requirements on latency, reliability,

and dynamism.

4.1 Devices in Power Generation

Power generation consists of large generation plants

(e.g., nuclear plants, fossil fuel plants) which are

capable of generating high voltage power. Those

plants work as systems that involve many compo-

nents such as generators, transformers, compressors,

and turbines. They are critical in power generation

and should be continuously monitored and reported

on their status. Beside those physical components,

there are also software components that are used for

remote control such as remote management system

(RMS). RMS controls a power grid with restrictive

time requirements in communication. For example,

data related to Breaker must be transmitted no longer

than 2 seconds after the event has occurred (Ericsson,

2010).

Another important component in power genera-

tion is wireless sensors which are used for monitoring

the health of the generation devices. They communi-

cate with each other to send data to detect any fault

in the generation process. Thus, reliable communi-

cation is critical. However, sensors are vulnerable

to harsh environmental conditions such as wind and

rain (US.DOE, 2004), which increases communica-

tion dynamism. Such a dynamism should be consid-

ered in designing the communication system in SG.

4.2 Devices in Power Transmission

Supervisory control and data acquisition (SCADA)

systems are widely used in power transmission for re-

mote monitor and control. A SCADA system con-

sists of different types of Intelligent Electronic De-

vices (IEDs) which work to control different parts in

the system. An IED has the capability of sending data

and interacting with other IEDs and with the control

center. An IED may have restrictive time constraints

on communication. For example, IEDs responsible

for substation protection and control have to trans-

mit their data within 12-20 ms (Ericsson, 2010). In

order to satisfy such a requirement, a reliable com-

munication protocol is required. Another example is

protection relays which are responsible for detecting

and overcoming failures in power devices. Protection

relays also have a restrictive time requirement that the

Tailoring DDS to Smart Grids for Improved Communication and Control

435

response must be within 3 ms to avoid faults that may

lead to blackouts (IEEE-Power-Engineering-Society,

2004; Schwarz, 2004).

Substations are another core components in power

grid. A substation has a microcomputer for remote

control and to communicate with other systems in

the grid, which require reliable data transmission in

a timely manner.

Phasor measurement unit (PMU) is a device for

measuring the health of the grid. PMU has the ca-

pability of processing and communicating data with

other devices in SG (Ek, 2014). PMU is required to

provide data to the control center at the rate of 6 –

60 samples per second (Rihan et al., 2011), which is

critical for safety control.

4.3 Devices in Power Distribution

The distribution domain in SG introduces the notion

of distributed energy resources (DER) such as so-

lar panel and wind turbines. Those systems contain

different components to generate power and commu-

nicate with other devices. Communication between

DERs and other components is a challenge due to un-

reliable environments where a large number of DERs

are envisioned to communicate via wireless connec-

tion and such a connection is vulnerable to interfer-

ence and harsh weather conditions (Yu et al., 2011).

DERs have a set of communication constraints to

be satisﬁed to guarantee the grid reliability. An ex-

ample is that the reading interval of meters has to be

5-15 minute (DNV.GL, 2014). This requires any DER

device to issue its data with respect to the speciﬁed in-

terval and any violation of the constraint may result in

unreliable data.

4.4 Devices in Power Consumption

The power consumption domain involves devices

such as smart meters and wireless energy moni-

tors. Smart meters communicate with power utili-

ties to send and receive data such as the total cost

of the electric power and pricing information in real-

time (DNV.GL, 2014), which requires continuous and

accurate updates on a regular basis.

5 TAILORING DDS

In this section, we describe how the DDS framework

can be tailored to satisfy the communication require-

ments identiﬁed in Section 4. We explain howthe reli-

able and restrictive time constraints can be addressed

by QoS in DDS. We also describe how to tailor the

discovery mechanism in DDS to satisfy the dynamism

requirements in SG communication.

The devices studied in Section 4 can be catego-

rized into publisher-related, subscriber-related, and

system-related in the context of DDS, depending on

their functionalities. Figure 3 shows the categorized

devices. In the ﬁgure, the rectangle shape represents

the components that can work as a system (with mul-

tiple components inside them) in the domain. The

oval shape represents publishers and the dashed rect-

angle representsthe componentsthat can work as both

publisher/subscriber. Sensors are categorized as pub-

lishers as they are used to sense and report on the

health of the devices in power generation. Smart me-

ters in the power consumption domain are categorized

as publishers/ subscribers as they are used to send

usage readings to utilities and receive from utilities

the total cost of the electric power corresponding to

the readings. On the other hand, SCADA systems

in the power transmission domain are categorized as

systems as they consist of multiple devices with data

sending and receiving functionalities (i.e., publishers/

subscribers).

Figure 3: Devices in the smart grid domain.

DDS requires its implementation to be capable

of scaling to a large number of subscribers (Object-

Managment-Group, 2014). It also requires the imple-

mentation to have a built-in discovery service that al-

lows publishers to dynamically discover the existence

of subscribers and vice-versa. The DDS speciﬁca-

tion also prescribes publishers and subscribers to be

able to set up QoS contracts at the time when their

intent to publish/subscribe data is declared (Object-

Managment-Group, 2014). Given that, we focus on

two aspects in tailoring DDS to SG – i) tailoring QoS

to satisfy reliability and restrictive time constraints

and ii) tailoring the discovery service to improve the

dynamism in SG.

5.1 Tailoring QoS

A signiﬁcant feature of DDS is the provision of QoS

on a per-entity base. This is important especially

in systems whose involved devices have different re-

SMARTGREENS 2016 - 5th International Conference on Smart Cities and Green ICT Systems

436

quirements the same quality concern (e.g, different la-

tency requirements). DDS does not only offer a set of

QoS, but also allows system developers to set differ-

ent QoS parameters for different entities such as Top-

ics, Data Readers, and Data Writers.

To satisfy time constraints and requirements for

reliable communication in SG, the QoS in the DDS

framework needs to be tailored. Different compo-

nents should be able to set different contracts. The

communication between publishers/ subscribers can-

not be established unless both parties agree on the set

of QoS. Given that, we suggest the following QoS at-

tributes to be tailored:

• Latency (Deadline): This attribute can be used

to set restrictive time constraints for components,

which allows publishers/subscribers to specify

how fast they can publish/receive data (Object-

Managment-Group, 2015). The attribute can be

tailored depending on the constraints of the com-

ponent on communication. For example, mission-

critical components (e.g., protection relays) in SG

must tailor this attribute to set up a rigid deadline

for data transmission.

• Latency Budget: This is another important qual-

ity attribute that can be used for implementing op-

timization on the publisher side to accommodate

the maximum acceptable latency of subscribers

which refers to the delay from the time the data

is written until the data is inserted into the sub-

scriber’s cache. It can be tailored to deﬁne com-

munication rules to satisfy restrictive time con-

straints for devices. One approach to tailoring this

attribute is to use it in conjunction with a priority-

based transport protocol to set up higher priority

for data with low latency budget.

• Reliability: This enables a Reader to receive data

reliably sent by a Writer. It can be used in SG

to satisfy reliability constraints. There are two

sub-parametersfor setting this QoS – Reliable and

Best Effort. The Reliable setting enforces reliable

data exchange. For example, for the components

(e.g., smart meters) that can tolerate a certain la-

tency to receive data, the Reliable parameter can

be set to retransmit data until received. On the

other hand, the Best Effort setting has no relia-

bility mechanism. Thus, it can be used for the

components that periodically publish data sam-

ples where only the latest one matters (e.g., data

sensors). Tailoring this quality attribute is carried

out by making appropriate setting for the given re-

liability constraints.

5.2 Tailoring Discovery Mechanism

The key for establishing communication between

publishers and subscribers is discovering the exis-

tence of participants. In the DDS context, this re-

quires DDS entities to be informed of each other’s

existence for communication. Not only this, the dis-

covery feature should also provide communication

guides between different publishers and subscribers.

For example, communication guides can be provided

as IP multicasting by devices where the DDS infras-

tructure manages the group membership.

The DDS speciﬁcation deﬁnes a Real-Time Pub-

lish Subscribe protocol (RTPS) which describes a dis-

covery functionality. The main purpose of RTPS is

to support the interoperability of applications built

upon different implementation platforms of vendors.

RTPS (Object-Managment-Group, 2014) deﬁnes two

discovery protocols – Participant Discovery Protocol

(PDP) and Endpoint Discovery Protocol (EDP). PDP

is used for discovering different domain participants,

while EDP is used for matching data readers and data

writers. However, these protocols are deﬁned in a

generic way, which may cause some drawbacks with

respect to system dynamism. For example, they re-

quire a large amount of periodic data exchange, which

is not applicable to the components that have limited

computing resources, processing, or power resources

(e.g., batteries). In the following, we describe features

with respect to DDS discovery service that need to be

tailored to improve SG dynamism.

• Providing Communication Guides for Differ-

ent Components: As mentioned in Section 4, SG

requires a dynamic communication system. Thus,

it is important to allow system components to be

added and removed in a dynamic way. For exam-

ple, when a new component (e.g., new publish-

ers/subscribers) is added to the system, the com-

ponent can be reached efﬁciently by the use of

discovery hints (e.g., the group membership pro-

vided by DDS).

• Enabling Multiple Discovery Strategies for

Different Components: The DDS speciﬁca-

tion allows the separation of its implementa-

tion from discovery services (Object-Managment-

Group, 2014). However, a DDS implementation

in SG should have a built-in discovery service to

support ﬂexible and dynamic discovery of com-

ponents. Components with limited resources may

not be able to handle heavy interactions. For

smooth integration of these components, the DDS

infrastructure must provide alternative discovery

mechanisms. For example, IEDs or sensors in SG

need a simple discovery protocol such as Simple

Tailoring DDS to Smart Grids for Improved Communication and Control

437

Service Discovery Protocol (SSDP). On the other

hand, for the components that use ﬁxed servers

and are capable of managing heavy interactions,

static discovery or ﬁle-based or server-based dis-

covery mechanisms can be used. It is worth men-

tioning that RTPS allows different implementa-

tions to support multiple discovery protocols (i.e.,

multiple implementations of PDPs and EDPs).

6 CONCLUSION

We have described identifying communication re-

quirements for SG and tailoring DDS to SG based on

the identiﬁed requirements. Requirements are iden-

tiﬁed for the devices involved in the four different

domains of SG. Given the identiﬁed requirements,

we suggested tailoring the QoS and discovery mech-

anism of DDS. We described tailoring latency, la-

tency budget, and reliability for QoS and tailoring the

features of providing communication guides and en-

abling multiple discovery strategies for the discovery

mechanism.

For the future work, we plan to investigate how

the tailoring points identiﬁed in this work should be

applied to the design speciﬁcation of DDS. In partic-

ular, we will look into Data Centric Publish/Subscribe

(DCPS) in DDS which is responsible for efﬁcient de-

livery of data between publishers and subscribers. We

also envision to study the data model that ﬁts both

DDS and SG to facilitate interoperability of involved

systems.

REFERENCES

A. Alkhawaja, L. Ferreira, M. A. Message oriented middle-

ware with qos support for smart grids. Conference on

Embedded Systems and Real Time, Caparica, Portu-

gal, pages 1 – 13.

Bakken, D., Schantz, R., and Tucker, R. (2009). Smart grid

communications:qos stovepipes or qos interoperabil-

ity. Technical Report TR-GS-013, Washington State

University.

Corsaro, A. (2014). The data distribution service tu-

torial. Technical Report 4.0, PrismTech, http://

creativecommons.org/licenses/by-sa/4.0.

DNV.GL (2014). A review of distributed energy re-

sources. Technical Report 6.1, DNV GL, https://

www.dnvgl.com/energy/.

Ek, E. B. (2014). Utilization of Phasor Measurement Unit

Measurements as Basis for Power System State Esti-

mation Interface. PhD thesis, Norwegian University

of Science and Technology, Norway.

Ericsson, G. N. (2010). Cyber security and power system

communicationessential parts of a smart grid infras-

tructure. IEEE Trs. on Power Delivery, 25(3):1501–

1507.

Fang, X., Misra, S., Xue, G., and Yang, D. (2012). Smart

grid the new and improved power grid: A sur-

vey. IEEE Communications Surveys and Tutorials,

14(4):944 – 980.

IEEE-Power-Engineering-Society (2004). IEEE standard

communication delivery time performance require-

ments for electric power substation automation. Tech-

nical Report 1646.

Lu, X., Lu, Z., Wang, W., and Ma, J. (2011). On network

performance evaluation toward the smart grid: A case

study of dnp3 over tcp/ip. Global Telecommunications

Conference, IEEE, Houston, TX, pages 1 – 6.

Modbus-IDA (2006). Modbus protocol speciﬁcation.

http://www.modbus.org/docs/Modbus

Application P/-

rotocol

V1 1b.pdf accessed: 12/02/2015.

NIST (2010). Nist framework and roadmap for smart grid

interoperability standards. Release 1.0, National Insti-

tute of Standards and Technology.

Object-Managment-Group (2014). The real-time publish-

subscribe protocol (RTPS) DDS interoperability wire

protocol speciﬁcation. Version 2.2 formal/2014-09-

01.

Object-Managment-Group (2015). Data Distribution Ser-

vice (DDS). Version 1.4 formal/2015-04-10.

Rihan, M., Ahmad, M., and Beg, M. (2011). Phasor mea-

surement units in the indian smart grid. Innovative

Smart Grid Technologies - India, 2011 IEEE PES,

pages 261 – 267.

Schwarz, K. (2004). IEC 61850 Communication Net-

works and Systems in Substations. 1.1 2004-03-

22, International Electrotechnical Commission, http://

www.iec.ch/smartgrid/standards/.

Twin Oaks Computing, I. (2011). What can dds do for you.

Technical Report 6, www.twinoakscomputing.com.

US.DOE (2004). Assessment study on sensors and automa-

tion in the industries of the futur. Technical report,

Dep. of Energy Efﬁciency and Renewable Energy.

World-Bank (2015). Electric power transmission and distri-

bution losses. Technical report, The World Bank.

Yu, F., Zhang, P., Xiao, W., and Choudhury, P. (2011). Com-

munication systems for grid integration of renewable

energy resources. IEEE Network, 25:22 – 29.

SMARTGREENS 2016 - 5th International Conference on Smart Cities and Green ICT Systems

438