A detailed explanation of RFID systems and case studies

　　We are increasingly encountering radio frequency identification (RFID) systems in our daily lives and work. From inventory control to fast checkout in supermarkets, this technology is transforming many existing applications and enabling new ones. On the frontend, the "signal chain" starts with small tags attached to units of interest; Tags transmit information in the form of a bit stream to RFID readers, which detect when tags are present in specific areas and read the information they carry. On the backend, server-based systems maintain and update the tag database, generate alerts within the enterprise, or initiate other information-based processes.

　　Most RFID readers currently use multiple processors to meet application requirements. Typically, the signal processor is connected to an analog-to-digital converter (ADC) and a digital-to-analog converter (DAC). Then, the network processor communicates with local or remote servers for information storage and retrieval. This article introduces how these seemingly completely different functions—signal conversion and network connectivity—are managed through a single processor in ADI's Blackfin processor series.

　　We will first briefly introduce RFID technology and discuss its potential for current and future applications. Next, we focus on RFID reader functionality, exploring the basic software components and server connections that need to run on RFID readers. Finally, some block diagrams provide some system configuration suggestions.

　　Today's applications and emerging applications

　　RFID technology enables many new types of applications by allowing simultaneous monitoring of multiple projects without the need for people to "touch" each item (such as handheld barcode scanners). Applications that can leverage this automatic identification include various fields such as inventory control, logistics management, monitoring, and billing.

　　Today, ubiquitous universal product codes for goods—such as '/em' (UPC)—are available as one-dimensional (1D) barcodes, which can meet almost every public purchasing need. Barcodes contain relevant information about the items they are associated with, which may include the project's suggested retail price and/or the location and date of manufacture. 1D and 2D barcodes can also be used to track detailed shipping details of items.

　　Barcodes work for individual items, but when many items need to be scanned, workflow efficiency decreases. For example, opening and scanning each item individually on a pallet containing hundreds or thousands of final products is impractical. But even if the scanned items are relatively small, such as groceries at the supermarket checkout, the correct alignment between the scanner and the scanned label must be established. More importantly, manipulating a large project to find barcodes can be challenging.

　　RFID technology replaces UPC with EPC (Electronic Product Code) in the form of bitstreams. At the very least, EPC allows automatic collection of the same type of information contained in barcodes and remote access, with minimal human intervention. Moreover, even if many identical items exist, the EPC can include more information related to the unique identifiers of the marked items. Moreover, unlike traditional barcodes, the orientation of the item or the ambient lighting conditions does not matter—it can still detect and track the item. Fog, darkness, and even warehouse dirt no longer mattered.

　　Here are more ways to use RFID systems:

　　In supermarket food trays and boxes, they allow for asset tracking and better asset pooling. By writing labels, additional information (such as sales dates) can be included. Additionally, automatic reordering can be implemented to maintain the correct inventory on shelves.

　　In the library, they can be used to automatically publish and return materials, which were previously individually labeled by barcodes to identify scanners.

　　On clothing labels, they can identify the true origin of the item. By using the label's identification number, the item can be authenticated or investigated as a forgery alone.

　　In the pharmaceutical industry, they can be used to prevent counterfeit and substandard goods.

　　In sports competitions, they can accurately track runners' progress during long runs.

　　Overview of RFID systems

　　RFID uses bitstream radio frequency (RF) transmission for communication, identifying, classifying, and/or tracking objects. Each object has its own RFID tag (also known as a repeater). The entire system uses a tag reader, a subsystem that receives RF energy from each tag. The reader has embedded software to manage the inquiry, decoding, and processing of received tag information; It communicates with storage systems that store tag databases and other related information. Figure 1 shows a conceptual diagram of the RFID system.

　　RFID readers

　　RFID readers provide connections between each tag and the tracking/management system. It comes in various shapes and sizes, usually small enough to be mounted on counters, tripods, or walls. Depending on application and operational conditions, there may be multiple readers who can fully serve specific areas. For example, in warehouses, reader networks can ensure that 100% of pallets are queried and recorded from point A to point B.

　　Overall, readers offer three main functions: bidirectional communication with tags to isolate individual tags; Initial processing of received information; and connect to servers that link information to the enterprise.

　　RFID readers must handle multiple tags within the field of interest—a crucial consideration in applications with many tags in confined space (for example, multiple labeled products residing on numerous factory pallets). )

　　The main challenge in multi-reader/tag scenarios is that conflicts occur when many readers query and multiple tags respond simultaneously. The most common way to avoid this problem is to use some form of time-division multiplexing algorithm. You can set the reader to query at different times, and the tag can be configured to respond after random intervals. It is clear that implementing this feature in embedded software provides additional flexibility.

　　RFID Transponder ("Tag")

　　RFID tags consist of an integrated-circuit (IC) chip that stores unique information about the tagged object (such as EPC data), an antenna (usually printed circuit patterns), a radio frequency energy received from the reader and transmits information, and a casing containing the tag components. It is worth remembering that the above term "object" can apply to any number of different things, from factory goods to animals to people. The distance from the tag to the reader is an important system variable and is directly affected by the labeling technology. Labels can be passive, active, or semi-active.

　　Passive tags

　　Passive tags are the simplest type. RF energy sent by the reader is specifically powered; they do not have integrated batteries, so they can be cheap, mechanically robust, and very small (for example, about the size of a thumbnail). However, passive tags have a limited reader-to-tag range because the received power depends on their physical proximity to the RFID reader.

　　The range of the link is also affected by the selected RF frequency. Low frequency (LF) tags typically use the 125-kHz to 135-kHz portion of the spectrum; Due to their limited range, they are mainly used for access control and animal marking. High-frequency (HF) tags mainly operate in the 13.56-MHz band, with a permissible range of several feet. They are typically used for simple one-to-one object reading, such as access control, charging, and tracking portable items like library books.

　　On the other hand, UHF tags operate at frequencies from 850 MHz to 950 MHz and have a fairly long range—10 feet or more. Additionally, because the available bandwidth may be wider, readers can query many of these labels at once, rather than performing one-to-one tag reading at lower frequencies. This feature helps minimize the need for multiple readers within a specific area, making UHF tags very popular in industrial applications for inventory tracking and control. However, UHF tags cannot effectively penetrate liquids, which is a major drawback, making them less useful for liquid-filled objects such as beverages and humans. To track these items, HF labels are commonly used.

　　In a 2004 passive label supplier survey, the price of UHF labels was expected to reach 16 cents per label in 2008, down from 57 cents in 2003—continuing to make labeling items a cost-effective method for asset and inventory tracking.

　　Semi-active labels

　　Like passive tags, semi-active tags reflect RF energy (rather than transmit) it back to the tag reader to send identification information. However, these tags also contain batteries that power their ICs. This allows for some interesting applications, such as when the tag contains sensors. In addition to static recognition data, each transponder can also transmit real-time attributes such as temperature, humidity, and timestamps. By powering simple ICs and sensors using only batteries—without including the emitter—semi-active tags achieve a trade-off between cost, size, and range.

　　Active tags

　　By using integrated batteries to power tag ICs (along with any sensors) and RF transmitters, active tags go a step further. Because they are self-powered, they can operate over a wider reader-to-tag range (up to 100 meters or more), which also means goods can pass through the reader faster than passive or semi-active tags. System. Additionally, active tags can carry more product information than EPC codes.

　　On the downside, batteries shorten the service life of active tags and increase their cost and size. Active tags typically operate in the 433 MHz and 2.4 GHz industrial, scientific, and medical (ISM) bands, which are available in most regions worldwide. Therefore, as more wireless consumer products appear in 2.4 GHz-based 802.11 and Bluetooth ® modules, the coexistence of active tags with these devices has become a significant issue.

　　Software architecture for RFID readers

　　After introducing the basic functions of RFID readers, we now consider how to implement the reader using the Blackfin converging processor. The three elements of RFID reader software architecture are: backend server interface, middleware, and frontend tag reader algorithms. Although different, all these elements of the software architecture can run simultaneously on a single Blackfin processor.

　　Backend servers and connections

　　Typically, RFID readers include a network component—wired, for example, Ethernet (IEEE 802.3), wireless Ethernet (IEEE 802.11a/b/g), or ZigBee ™ (IEEE 802.15.4)—which connects a single RFID reading event to a central server. A central server runs database applications with functions including matching, tracking, and storage. Many applications also have "alert" functions (triggers for reordering supply chain and inventory management systems, or alarm alerts for security applications).

　　By the way, readers are building around high-performance embedded processors running μClinux (also uClinux), which have clear advantages compared to those that do not have when communicating with backend servers. The presence of a powerful TCP/IP stack and the availability of SQL database engines greatly reduce the main integration burden during development.

　　Middleware

　　The term middleware used in RFID has some different definitions compared to its use in other embedded systems. In terms of RFID, middleware acts as the software conversion layer between the front-end RFID reader and the back-end enterprise system. Middleware filters data from the reader and ensures it does not read multiple times or bad data. In early RFID systems, middleware ran on servers, but now RFID data filtering usually happens on readers before being sent through the enterprise network. This added functionality is another advantage that embedded processors bring to this application space.

　　The front end of the reader

　　The system's filtering and transformation-intensive signal processing occur at the end of the front reader, requiring devices with strong signal processing performance typically associated with Blackfin processors.

　　A/D and D/A Converters: Now that we have the general meaning of RFID system components, let's focus on connectivity from the perspective of RFID readers. To communicate with tags, mixed-signal front-end (MxFE ®) ICs form an interface of interest.

　　The MxFE device is a universal midrange subsystem, including A/D and D/A converters, low-noise amplifiers, mixers, AGC circuits, and programmable filters. The output stream of I&Q data is directly connected to the processor's parallel port. ADI's MxFE IC series products form the highest-performing narrowband receivers, making them ideal for RFID and other applications.

　　Figure 2 shows a block diagram of a typical MxFE device.

　　Blackfin processors for RFID applications

　　Blackfin processors provide connectivity to both wired and wireless networks. Some processors (such as ADSP-BF536 and ADSP-BF537) have 10 Base-T / 100-Base-T Ethernet MACs on the chip. On the wireless side, all Blackfin processors can connect directly to 802.15.4 ZigBee and IEEE 802.11 chipsets via SPI ® and SPORT peripherals. It can achieve line-speed transmission without consuming the entire processor bandwidth.

　　Additionally, Blackfin processors include parallel peripheral interfaces (PPIs), which can be directly connected to ADCs and DACs, as mentioned above. Some Blackfin processors include two PPIs that can further expand system functions—for example, allowing cameras to be connected to RFID readers. Beyond RFID applications, these Blackfin features are especially attractive for 1D and 2D barcode applications, as Blackfin can perform system control, networking, and image processing on the same device.

　　For RFID applications, the single way RFID readers query the tag is usually sufficient for PPI. First, the PPI is configured in transmission mode, with the processor sending the digital sequence to the DAC. The transmitted sequence is converted into an analog signal, then up-converted and sent out to excite/wake the local RFID tag, followed by a response. Meanwhile, PPI is reconfigured as a receiver in a small number of processor system clock pulses (see EE-Note 236), as shown in Figure 3. This way, the down-conversion RF signal can be sampled by the ADC and directly into Blackfin. In this diagram, the time between each receive (Rx) and send (Tx) interval is measured in the system clock cycle. The time elapsed allows the transmitted signal to reach the tag and the tag to transmit the response.

　　In some RFID applications, the Blackfin processor itself can serve as a server—for example, when big data storage and database operations are not needed. For example, imagine an elderly parent wearing a bracelet with a tag, which can be monitored inside the house. If no signs of activity are found within the specified intervals, monitoring agencies can alert registered friends or relatives.

　　The software components that make up Blackfin RFID reader infrastructure can be found on the Blackfin.uClinux.org website. The product includes drivers required for interfaces with mixed-signal front-end ICs, as well as DMA drivers that are very useful when moving data through the system. A μClinux-based network stack and SQL database engine are also available. From a system perspective, other features (such as the 802.11 Wi-Fi card, USB thumb drive, and CompactFlash card interface) can integrate with Blackfin devices very quickly. For more information, please refer to http://blackfin.uclinux.org.

　　Example of an RFID system

　　Wired RFID system

　　The most common application of RFID is asset management, which can track pallet movement within warehouses by reducing inventory loss, eliminating delivery errors, improving distribution logistics, and minimizing stockouts. RFID systems in large warehouses can track the movement of pallets loaded with goods from entering the warehouse to leaving. Such systems rely on fixed RFID readers throughout the warehouse and at inbound/outbound transport points.

　　As a means to simplify wired infrastructure, Power over Ethernet (PoE) networks are the ideal choice for these types of applications. IEEE 802.3a/f PoE handles network systems in low-power applications. The PoE system (as shown in Figure 4) consists of power supply equipment (PSE) and power supply equipment (PD). PSE supplies power to the Ethernet lines, while PD (for this purpose) converges the network processor and its surrounding components. PoE recommends a maximum cable length of 100 meters, suitable for many embedded RFID applications because of its relative mobility and eliminating the costs associated with installing traditional AC cabling and sockets.

　　In addition to RFID acquisition software, network processors supporting embedded RFID applications also require sufficient performance and integration to handle complex multi-layer IP stacks. The ADSP-BF537 Blackfin processor—including a 10-Base-T / 100-Base-T Ethernet MAC—is a great example of this integration. For example, many Ethernet PHY devices provide state pins with the ability to interrupt when state changes. This feature is seamlessly integrated with Blackfin interrupt functionality, generating a powerful, low-power system.

　　Low-cost wireless RFID

　　Handheld scanners suitable for applications such as forklift scanners or portable devices cannot perform wired or PoE operations. Wireless protocols like IEEE 802.11b/g allow RFID readers to connect to wireless access points, as shown in Figure 5. Blackfin processors can be connected via serial or parallel interfaces to the 802.11 chipset. Additionally, due to their computing power, these processors support both separate MAC and full MAC 802.11a/b/g implementations. For example, system integration of a CompactFlash 802.11b card may require a full MAC interfaced via Blackfin's asynchronous memory port. Split MAC implementations usually use SPORT or SPI interfaces—the lower MAC resides on the wireless chipset, while the upper MAC runs within the Blackfin software.

　　While its stack and processing requirements can be easily handled on single-core processors, wireless applications are testing the boundaries between performance and power consumption. Dynamic power management features using low-cost convergence processors (such as the ADSP-BF531) enable power management and provide scalable performance according to application requirements. These dynamic power consumption modes are designed to provide flexible performance and power configurations for almost any network system.

　　High-performance systems

　　In emerging applications, RFID technology pairs with other devices, such as biometric sensors or CMOS image sensors. As shown in Figure 6, in advanced applications of security authorization and personnel access control, RFID is combined with image analysis to ensure that in a secure environment, not only are there N people in the room, but all are "authorized personnel."

　　The computing demands of these applications are very suitable for handling dual-core fusion processors, such as the ADSP-BF561. Additional processor cores not only effectively double the computing load the device can handle; It also offers some surprising structural advantages, which are not very obvious.

　　Traditionally, dual-core processors use discrete and often distinct tasks running on each core. For example, a single core may perform all control-related tasks—such as networking, interfaces with large-capacity storage, RFID acquisition, and overall flow control. This core is also where the operating system or kernel might reside. Meanwhile, the second core can be dedicated to the application's high-intensity processing capabilities. For example, the video processing part of a human recognition algorithm might run on the second core, and the resulting packets might be passed to the first core for transmission via network interfaces.

　　The dual-core ADSP-BF561 includes dual high-speed L1 instruction and data memories (each local), as well as shared L2 memory between the two cores. Each core can equally access various peripherals—video ports, serial ports, timers, and so on. As mentioned above, one core of the ADSP-BF561 manages RFID acquisition and network components, while the other core can be dedicated to an image classification system capable of real-time detection, classification, and tracking of objects.

　　μClinux

　　The μClinux operating system is a popular choice that facilitates network connectivity—the largest software component in card readers—as well as the key requirements for robustness and standards compliance. When reading RFID tags, it is essential to ensure real-time requirements are met. Since the μClinix scheduler is not strictly real-time, it can be replaced with the ADEOS real-time scheduler, which can securely block μClinux interrupts until the critical real-time processing is completed. This means front-end card reader software can execute in real time from the ADEOS domain, while middleware and backend server interfaces can run in traditional μClinux environments. This division provides users with hard real-time control over their applications while allowing access to all the benefits of open-source software. For more information about μClinux or ADEOS, please refer to the BlackfinμClinuxWiki.

　　Figure 7 shows an ADI MxFE evaluation board connected to the Blackfin ADSP-BF537 STAMP development platform, which runs MxFE driver code, μClinux operating system, and TCP/IP network stack.

　　Conclusion

　　As we have shown, RFID applications no longer require dedicated signal processors for ADC/DAC interfaces and microcontrollers for networks. The Blackfin series fusion processors can handle networking and control, offering sufficient performance for converter interfaces and pattern matching algorithms. In turn, this can bring lower-cost bills of materials and faster time-to-market for the next wave of RFID applications.