Create an RFID-Ready Architecture
Turning massive amounts of data into operational intelligence means addressing the new challenges that radio frequency identification and sensors present.
by Peggy Chen

June 15, 2005

Many organizations that are focused on the rush to meet radio frequency identification (RFID) compliance deadlines—Wal-Mart, Metro Group, the United States Department of Defense, Best Buy, and Target, to name a few—are deploying RFID slap-and-ship pilots. Although a standalone slap-and-ship system might be the fastest way to meet RFID mandates and the most economical short-term solution, if you don't consider architectural and integration issues, you might find yourself dealing with an integration nightmare with multiple silos of fragmented information.

You will undoubtedly be collecting more and more data once RFID and other sensors are deployed in your organization. All your enterprise applications will ultimately integrate and leverage this data. You must consider the impact of RFID across your entire technology stack, not just the locale where RFID readers are being deployed (such as the warehouse).

Consider the analogous case of bar codes. The use of bar codes is now widespread—from supermarket and department store items to self-service postage labels, return material authorization notices, and tax returns. And, software applications and infrastructure have evolved to support them. Whereas bar codes usually identify the type of product, RFID tags uniquely identify a particular asset or item. The ramifications of RFID tags on enterprise software infrastructures are bound to be much wider than that of bar codes.

Therefore, you need to understand how the flow of RFID and product data will work throughout your enterprise systems, and ensure that your architecture will be ready to support these new sensor-based computing and event-driven paradigms as requirements for new applications emerge. This article takes a look at some key challenges RFID presents and discusses how you can turn them into a sustainable advantage for business and IT.

Capture Data From the Physical World
Increasing visibility into your supply chain depends on an accurate view of where your assets have been, where your assets are now, and where your assets are going. You also need to know the condition, status, and state of every asset. Getting a comprehensive view of what is happening in the physical world of your organization requires capturing data from not only RFID readers but also other types of sensors, including temperature, humidity, shock, location, and more.

For example, consider the shipping of pharmaceutical drugs or perishable goods that have to be kept at a certain temperature. Tagging such medications and foods with RFID tags and temperature sensors enables constant monitoring. Tagging these shipments with active Global Positioning Systems (GPSs) enables them to be definitively located. The data from all the various tags and devices could be transmitted across wireless networks using wireless fidelity (WiFi), Global System for Mobile Communications (GSM), and other technologies, including peer-to-peer networks. Collecting this data and processing the information has its own unique challenges.

The challenge for the IT department is building an architecture that has the flexibility to integrate with any sensor or device, as well as capabilities to filter events at the source, route events to applications that need to consume them, and provide generic event-processing capabilities for processing streams of RFID or sensor data into business events. The new breed of RFID edge middleware that provides device abstraction and event filtering needs to be well integrated into existing enterprise architectures (EAs). This provides enterprises with the most flexibility for developing and deploying RFID and sensor-enabled applications.

Manage Devices on the Network
True visibility into your supply chain will require connection to an assortment of devices in the warehouses and manufacturing plants, including RFID readers, RFID printers, response devices, material handling equipment, and more. As these devices come online and get connected to your IT systems, how will you manage the wealth of devices on the network? How will you ensure that these devices—like the rest of the servers on your network—are live, healthy, and operational?

The RFID and sensor-based computing solution you deploy in the manufacturing facility, warehouse, or distribution center needs to be managed without an administrator on site. Remote-device management becomes even more important when you look toward multisite and potentially global deployments. Solutions that monitor and manage software elements, Web applications, hosts, and the network now need to also drill down to the edge servers deployed to provide a central view of all devices and sensors on the network collecting data from the physical world. These centralized management solutions need to help your system administrators resolve issues with devices and the networks that support collection of RFID information, enable root-cause analysis, and facilitate general management and administration.

As organizations move beyond the basic RFID pilot, whether scaling out to more distribution centers and warehouses or integrating RFID into the manufacturing line, the amount of RFID and sensor data captured will grow exponentially. Although today's data volume might be low, IT departments must be prepared to provide a middleware and data-management solution that can scale with the volume of information. To ensure cost-effective scaling of this infrastructure in response to increasing demands, many vendors are looking at grid-based infrastructures.

Making sense of all the Electronic Product Code (EPC) numbers being read at every dock door in all your warehouses and distribution centers requires correlating EPC numbers with product information. Storing both RFID and sensor data in a centralized data hub provides the most accurate and flexible view of "reality," which can then be integrated into all the applications that depend on this data. An EA with user-definable rules for processing the plethora of event sightings into actionable business interactions will provide the best foundation for transforming and optimizing your supply chain business operations.

The key to identifying the black holes in your supply chain is analyzing the reality that RFID and sensor data provide. To effectively monitor and control business operations, managers need to be able to quickly assemble and analyze a contextual view of any situation—taking into account both historical and current information such as objectives, past performance, and external forces. They also need to be able to share their findings with other people throughout their organization and either take action directly or delegate action.

Delivering increased visibility, insight, and intelligence about the organization's operations requires an EA that not only captures the events from the physical world and translates them into business events, but can also perform complex calculations and analysis on composite business events. The mountains of RFID data collected are only useful when business users can visualize that data in context through reports and rich realtime visualization capabilities that make the data actionable by users.

Business intelligence (BI) and Business Activity Monitoring (BAM) capabilities are critical to delivering the right information to the right user at the right time. As such, EAs need solutions that can identify, model, and apply statistical models to trends within their operational environment, as well as anticipate and plan for environmental or business practice changes.

Define New Business Processes
RFID holds the potential of transforming business processes. But how do you make that become a reality? Although applications could directly consume and act on RFID data, this could create lock-in and limit your flexibility to update particular business processes as they change over time. How can you maximize the benefits of sensor technology with minimal applications impact? Further, how do you enable iterative business process enhancements and ensure that all applications and third parties that need the data have access to it?

To realize the impact of RFID on your business processes, you need an EA that can translate raw RFID data into one or more business events and compose them into appropriate business applications. Complex event processing enables you to process multiple streams of raw events, such as RFID or sensor reads, into complex events, which have significant business meaning and can then be dispatched to the appropriate application.

Business process modeling tools based on standards, such as Business Process Execution Language (BPEL) for Web Services, will allow you to assemble a set of discrete services into an end-to-end process, integrating not only RFID data but also data from your enterprise systems. BAM capabilities enable you to detect, correlate, and analyze separate events, such as RFID scans, bar-code readings, and events out of an enterprise application or a supply chain process event.

Furthermore, a business-to-business (B2B) solution allows you to extend these business processes beyond your four walls by providing the ability to define, configure, manage, and monitor the exchange of information electronically between organizations. Collectively, integration, process modeling, and monitoring capabilities and an event-driven architecture ensure that enterprises can quickly respond to changing business dynamics.

Build on a Service-Oriented Architecture
Many people think that RFID is just an issue for the operations people within an organization. However, once you consider the IT impact of deploying RFID readers across all your facilities, it becomes apparent that RFID is actually a cross-functional exercise that your IT department and enterprise architects also need to address.

Organizations need the freedom to implement RFID in increments and the flexibility to re-engineer processes that are RFID-dependent. But to incorporate RFID into your EA, you don't need to invest in a new architecture. Service-oriented architecture (SOA) features both services and events. RFID generates events that fit directly into an SOA.

In fact, the infrastructure services in an SOA environment (routing, transformation, process automation, complex event processing) enable integration of RFID data into existing or new applications and IT systems, so you can adjust more quickly to meet changing business needs. Furthermore, an SOA deployed on a grid infrastructure enables organizations to start small while planning for the long term, leveraging reduced integration complexity at a lower total cost (see Figure 1).

Let me illustrate this RFID-enabled EA in the context of a large electronics retailer, Good Circuits. Good Circuits wants to track and trace shipments as they arrive at its distribution centers with the long-term goal of tracking cases and pallets as they move within the company supply chain from distribution centers to back rooms of stores and then onto shelves. Good Circuits will start small by automating one business process—shipment receipt and reconciliation—and will deploy this solution at two distribution centers. Initially, pallets and cases coming from their 10 largest suppliers will be tagged with RFID. This will allow the company to build the right architecture for the long term while delivering business value from RFID in the short term.

The key principles that Good Circuits is applying to meet its business objectives include:

• Leverage existing systems using SOAs.
• Design an extensible platform that can be used to build a real-time supply chain visibility solution.
• Keep the initial footprint small to experiment and learn from the first phase.

The software architecture shown in Figure 1 enables Good Circuits to achieve these objectives: The SOA infrastructure in the application server enables the company to leverage existing investments in software and hardware; the application server capabilities enable the company to deploy new applications that leverage events, event processing, routing, and analytics; and the use of common services such as these enables the company to keep its development costs low. Good Circuits will use standards-based technologies, such as BPEL and Web Services standards (WS-Security, WS-Reliability, and so on) for SOA (implemented within the application server) and EPC standards for RFID, to ensure that it's well-positioned for the future.

Maximizing RFID Benefits With Minimal IT Impact
The adoption of RFID technology today allows organizations to stay ahead of the curve and achieve a certain level of business benefit. But turning the promise of RFID into a sustainable competitive advantage depends on how effectively you turn the vast amounts of collected data into operational intelligence. Mountains of data, encompassing all aspects of the business value chain, will push the boundaries of information management in terms of scalability, reliability, and security. The key value to capturing this data will be the ability to impact your business processes across your entire supply chain without creating information silos to address separate applications.

To protect your initial investments and realize maximum long-term returns, you need to ensure that you base your RFID strategy on a sound architecture—an architecture that adapts to changes in technology, standards, and business dynamics. Such an architecture needs to ensure that all your business constituents benefit from what RFID has to offer. By deploying RFID-enabled business processes in your SOA, you have the ability to enhance and optimize your RFID solution iteratively while imposing minimal to no impact on existing systems.

In building your architecture, focus your IT dollars on building applications that deliver business value, rather than integrating lots of middleware. Take a platform-based approach that integrates a complete sensor-based computing infrastructure for capturing, managing, analyzing, accessing, and responding to RFID and other sensor data implemented in the physical environment.

Make sure your architecture includes a framework for plugging in and managing RFID devices; the data hubs for centralized data management; and integration services to develop composite business processes and enable quicker communication with internal and external systems, customers, and partners. To make sense of the plethora of information, ensure that you have BI and BAM tools to analyze the data, and development tools to enable the development of rich applications for desktop, mobile, wireless, voice, and portal applications. Also ensure that your platform includes enterprise management for managing your entire enterprise infrastructure, and grid technology to provide the foundation for the most scalable and reliable RFID middleware platform.

Wherever you are in the RFID adoption cycle, plan for the future and start architecting for your long-term vision now. Building on the right EA will take you one step closer to achieving your objectives.

About the Author
Peggy Chen is group product manager of RFID and Sensor-Based Services at Oracle. She is responsible for cross-product solution strategies and product management. Prior to this position, Peggy oversaw product management of Oracle's mobile, wireless, and voice solutions, including Oracle Application Server Wireless and Mobile E-Business Suite.