By Donna Imam and Mike Dow

Abstract: Two way communications is at the heart of the smart grid but just as pagers and land lines are becoming a thing of the past, so will the ways with which we exchange information today. The real key to a ‘Smart’ grid is being able to disassociate the communication technologies, from the networking engine so that the system is flexible enough to use the communications medium best suited for the geographical area and the application. By stabilizing the networking backbone, the rework of communication software and hardware can be minimized as technologies, the geographic area, or the regulatory needs change. One way to solve the challenge of interoperability between various standards for communications is to allow physical media independence by the adoption of Internet Protocol (IP) networking for both wired and wireless smart object communications. This paper will discuss how to approach the system architecture so to provide the maximum flexibility and backward compatibility for the smart grid as it is deployed in steps.

Standards Abound: Look anywhere and you will hear about smart grid and smart meter standards developing, changing and becoming ratified. For a long time power engineering has been on the backburner for developed countries. The grid has been in place for years and things were running like a well oiled machine. With the increase in demand for energy world wide one of the biggest infrastructure changes is happening right before us and connectivity is the face of the grid’s future. With change comes challenges and to meet these engineers and utilities must weigh the tradeoffs that come with the best technical solution versus the economic or pragmatic one.

Smart home connectivity to the smart grid

Without seamless connectivity paired with command and control between the home and the utility there can be no smart grid. The challenge that engineers face today is many faceted. Not only must each high energy consuming appliance in the home be managed without interrupting the flow of life but this exchange of information must be done securely and fast enough for real time demand control. These problems have been solved before with the debut of the internet and cellular communications but with every new medium of communication there are a different set of challenges.

The biggest challenge for communication today is the diverse ways by which utilities can access usage data and enact demand response. This poses a complicated scenario of how to collect data from devices housed in various geographies while using the best possible way to address the problem. The snag is solving cost effectiveness without sacrificing requirements. The rush is implementing solutions today that make sense for the future. As utilities grapple with the reality of having to upgrade hardware (HW) in the field more often then 15 years; vendors fight to prove that they have the crystal ball solution. But is that really possible?

The historical perspective on communications protocols: Historically, communication stacks have been top to bottom vertically integrated and single purpose. For instance, take Bluetooth®, ZigBee®, ZigBee® Pro or WiFi ™ protocol stacks. In each of these stacks the OSI layer functionalities are specified from the PHY to the Application layer profiles. Interoperability between devices was paramount and the key value proposition for the organization. Initially Bluetooth and ZigBee targeted their stacks at broad horizontal markets with little success. They made the mistake of trying to be all things to all people. Bluetooth did not really start taking off until the wireless cell phone headset usage gained traction. To this day, that remains the primary use of Bluetooth. ZigBee initially had the same approach and quickly found that it was not robust enough for large wireless node deployments, thus, ZigBee Pro was born. Heavy duty industrial vendors quickly realized that even ZigBee Pro did not have the reliability they required, and the protocols ISA100™ and WirelessHART™ were the result.

Upon the realization by ZigBee and Bluetooth SIG that they could not address every wireless application without regard for what the application aimed to accomplish they began to spawn totally new bottom-to-top stacks targeted at specific vertical markets. However, this broke with their interoperability value proposition, so, now they generally consider themselves “umbrella” organizations with a broader connectivity solution theme. For example, Zigbee RF4CE is under the ZigBee umbrella and targeted at RF remote controls for consumer electronics. Bluetooth Low Energy is targeted at wearable sensors connected to the watch or a cell phone.

Unfortunately this umbrella strategy is still limiting in that it creates stacks that are highly specialized and targeted. This is great for a specific vertical application, but, very limited in its application base. In both the broad horizontal and tight vertical stack, the physical layer is still specified to be of one type. This limits potential use of the communications protocol depending on geographic area or application.

 Diversify ways to connect physically: Connectivity to the grid can be achieved in many ways both wired and wireless. In the home area network (HAN) ZigBee IP has emerged as the preferred protocol in the United States.  However 802.15.4-2.4GHz radios are losing favor in the European market because the buildings there are typically of stone or brick construction instead of wood frame. 2.4GHz has a much harder time penetrating walls than sub-GHz radios. We are now seeing interest in a sub-GHz ZigBee stack for the European market. Looking at the US metering market, the wire mesh of a stucco coated house acts as a Faraday cage preventing RF signals from the meter from penetrating into the house to control appliances. Because of this and other use cases, power line communications such as HomePlug® Green PHY will be needed for approximately 10-15% of the HAN applications for smart metering. Other wireless methods such as Direct Wi-Fi, M-bus are also in the running for the HAN.

 Although proprietary sub-gig is predominant in North America, for the neighborhood area network (NAN), PLC, WiMax, Cellular/GRRS are also gaining mind share rapidly.

 For a device manufacturer, trying to develop a hardware/software connectivity strategy, all of these communication options are causing either complex designs to incorporate multiple protocols, or, limiting the design to a particular use case or application. Neither situation is ideal. So the standards war wages on.

 In the end utilities need a cost effective way to access usage data and interact in real time with individual appliances. They need an easy way to deploy meters with the right connectivity physical layer for the landscape that connect into their data management system without the need to define interoperability. The key to solving the smart grid connectivity dilemma is to embrace this diversity. How can one do that?

 The world is converging on IP connectivity in the middle: The hope on the horizon is that the “IP of Things” seems to be gaining traction world wide. The thought is that smart objects will have an IP address and will be directly addressable by standard IP protocol tools. In other words, from anywhere on the World-Wide-Web you could “ping” your toaster and get a response. What has really made this method viable is the introduction of several new IETF (Internet Engineering Task Force) standards.

 IP addressability doesn’t come without hurdles. It has been estimated that the world will run out of IPv4 addresses within the next 2-4 years. The IPv6 standard was recently introduced which extends the IP addressing from 32 bits to 128 allowing for 3.4×10³⁸ addresses. That’s a lot of smart nodes. The second hurdle that had to be crossed was the size of the IP header. Since IP was originally designed for wired networks, the header is quite large. Many sensor networks are battery powered and sending around a huge IP header would quickly drain the battery. IETF addresses this header issue with the  6LoWPAN standard. 6LoWPAN defines a header compression scheme for sending IPv6 headers over an 802.15.4 network. Additionally ROLL (Routing Over Low power Lossy networks) a work Group of IETF intends to add an IP mesh networking scheme called RPL (Routing Protocol for Low power lossy networks) to the IPv6 tool box.

 Figure 1 below is a diagram of what an IP centric stack would look like. Because of the many various needs at the physical layer, there is still quite a bit of instability at the PHY/MAC layers. There will always be new and better ways to transmit data, so, it is likely that the PHY/MAC layers will experience continual churn in perpetuity. At the Session, Presentation, and Application layers of the stack there will be as many variations as there are vertical applications. Having the flexibility at the upper layers of the stack is essential to the success of IP driven connectivity in more than one vertical market. Vendors are able to cater their upper level stacks to address Smart Energy 2.0 for HAN, stay with their proprietary protocol or implement profiles suitable for other vertical markets such as healthcare. But, once an application profile is developed it should be fairly stable as the targeted application is not likely to vary much over time.

Figure 1

These IETF standards are intentionally designed to be physical transport independent. Currently IP is both a wired and wireless network/transport protocol. With the addition of IPv6, 6LoWPAN, and RPL to the IP toolbox, IP based sensor networks are now viable. It is possible to begin building a communications stack strategy that leverages the stability of IP in the middle.

No Clear Winner: When it comes to the way to connect there is no clear winner. Table 1 below shows a snapshot of all the communications standards in play today and illustrates that the PHY will remain a diverse playing ground. What complicates the matter more is the preference for one way over the other. The preference is not just technology driven but also dictated by the most appropriate way to connect based on the needs of the locality (rural vs. cosmopolitan) and government regulations world wide policing the wireless space.

Because of the mandate for the new smart grid in the US to be IP based, there is some very significant progress being made towards an IP centric communication stack. The Smart Energy Profile 2.0 working Group which is a joint effort between the ZigBee Alliance and the HomePlug Alliance is creating what is expected to be the first IPv6 mesh network standard that is PHY/MAC independent. WiFi Alliance has also recently formed a similar alliance with the ZigBee Alliance to create a Smart Energy 2.0 profile that will work over WiFi. So for the first time, there is a communication stack that is being created that is intended to be IP based and will be capable of running over at least 3 different physical layers.

 So if you are device manufacturer building a consumer smart energy service gateway for a home, you have some clear simplifications of your hardware and software design that will minimize cost and will allow your gateway to easily support multiple transports depending on the application.

 If you are a connectivity solutions provider for consumer and home appliances the way you architect your solution to address maximum flexibility in communications around the world is still a major challenge. How many variations of the design are you willing to manage? How many SW solutions are you willing to maintain in-house and update periodically? Soon the R&D expense adds up and the strength in numbers of units becomes manufactured is diminished by having to manage too many variations and customers wondering if a washer dryer purchase will see them through for decades to come. 



Figure 2

Figure 2 above illustrates a hardware/software architecture that allows a single MCU/MPU plus multiple PHY/MAC modems to implement a single data pipe for smart energy data. By putting the stable part of the communication stack on an MCU/MPU, the system engineer can stabilize the majority of his design on the MCU/MPU platform of his choice while using the development tools he prefers. The PHY/MAC churn is now separated from his MCU/MPU choice. This now allows the systems architect to choose his physical modem or modems based on the application or geographical needs. For the US market, with an MCU/MPU plus a WiFi, 802.15.4, and HomePlug Green Phy modem, a smart energy gateway designer could effectively cover all possible US HAN needs. Furthermore, the basic IP toolbox could be repurposed and bolted on top of the new 802.15.4g “Last Mile” radio of his choice to implement the majority of a “Last Mile” communication stack.

Counter arguments to the multi-chip and IP centric solutions: System vendors can argue that SoCs with MCU/MPU with on ship Phys are the right solution for high volume applications. There is no doubt that the economies of scale will dictate that route for many applications once the standards are widely adopted and device manufactures can depend on one standard. We’ve seen this happen before with VHS technology, but VCDs, DVDs and now Blu-ray™ players have illustrated the rate at  which change in adoption of technology can occur.

Then there are network layer timing issues that can be sited to drive the need for a tightly coupled controller and Phy. In reality the Network layer is decoupled from the timing requirements of the PHY by the MAC and the MAC layer is where timing issues are handled. So as long as the MAC sits next to the Phy timing issues can be easily addressed. We can use the HAN as an example to drive this point home, in the HAN we see that a wired solution such as the HomePlug GP needs the SE 2.0 Network layer just s a wireless 802.15.4 does. Illustrating that for a vendor wanting to bring both solutions to market the optimal implementation is to house the network layer on the MCU/MPU to service both 802.15.4 (wireless) and HomePlug GP (wired).

In the end both scenarios will exist but the markets will continue to drive change and vendors towards flexible solutions. Embracing flexibility is one way to get the grid smarter; faster.

References:

 http://en.wikipedia.org/wiki/IPv4_address_exhaustion

http://www.redorbit.com/news/technology/1863775/internet_running_out_of_ip_addresses/index.html
http://www.wisegeek.com/are-we-really-running-out-of-ip-addresses.htm
http://www.potaroo.net/tools/ipv4/index.html
NIST Special Publication 1108: “NIST Framework and Roadmap for Smart Grid Interoperability Standards, Release 1.0”
6LoWPAN: Incorporating IEEE 802.15.4 into the IP architecture Internet Protocol for Smart Objects (IPSO) Alliance
IP for Smart Objects Internet Protocol for Smart Objects (IPSO) Alliance

Author Affiliation:

Donna Imam is Global Product Manager, Microcontroller Solutions Group at Freescale Semiconductor.

Mike Dow is Business Development Manager, Microcontroller Solutions Group at Freescale Semiconductor.