RFID Technology and Testing

As the cost for passive tags drops thanks to advances in submicron Complementary Metal Oxide Semiconductors (CMOS), the use of RFID for inventory applications is becoming nearly universal. Many experts believe the 96-bit Electronic Product Code (EPC), as shown in Figure 1, will be the next generation of the Universal Product Code (UPC), the familiar General Trade Identification Number (GTIN) imprinted in the barcode on a majority of products sold today. The varying applications of EPC RFID tags have moved the industry to classify the basic types of RFID devices, ranging from 1 to 5 according to the tag's read/write capability and passive or active power source.

RFID overview and design challenges
The passive class 1 tag in the 900 MHz and 2.45 GHz frequency ranges is ideal for many high-volume applications. The high frequency allows the interrogator to read the tag with a directional antenna for a greater communication range. Passive tags at higher frequencies also work with smaller, less complicated antennas making them more suitable for consumer applications.

Reading passive tags is somewhat different than the traditional full duplex data link. Unlike traditional active data links, the passive tag relies on the RF energy it receives to power the tag. Passive tags modulate some of the energy being transmitted by the interrogator to the tag in a process known as backscattering, as show in Figure 2. By changing the loading of the antenna from absorptive to reflective, a Continuous Wave (CW) signal from the interrogator can be modulated.

Passive tag readers are typically configured as a homodyne or single frequency conversion receiver as shown in Figure 3. A precision frequency source in the interrogator generates both the transmitter signal and the local oscillator for the reader's receiver. The unique homodyne architecture of the Class 1 RFID system presents some unusual challenges for the engineer. The backscattered modulation is far weaker than the CW signal from the reader's transmitter used to power the tag during backscattering. At baseband in the reader's receiver, the CW leakage translates to a large DC offset that can saturate sensitive amplifiers and digitizers.

Another challenge with the passive tag RFID system is the powering of the tag from received RF energy. Even though submicron CMOS requires very little power to operate, at a range of only a few meters very little power (- 10 to -15 dBm) is available. Complicating matters further, regulatory bodies worldwide have different maximum Effective Isotropic Radiated Power (EIRP) limits.

Since the uplink from the Tag (T) to the Reader (R) (denoted T=>R) is modulated from the interrogator's CW signal, it is possible to use spread spectrum techniques such as frequency hopping. Any spreading on the interrogator's signal will automatically be removed in the homodyne down conversion of the receiver since it shares the same Local Oscillator (LO) signal. After down conversion the interrogator's homodyne receiver has separated In phase (I) and Quadrature phase (Q) signals. The down-converted base-band signal is then digitized with Analog to Digital Converters (ADC) and digitally processed to determine the tag's ID.

Modulation and coding
RFID systems usually use simple-to-produce modulation techniques and coding schemes that lead to some design tradeoffs. A typical example is ISO 18000 Type C (also known as EPC Gen2, Class 1) which calls for Double Side Band-Amplitude Shift Keying (DSB-ASK), Single Side Band-ASK (SSB-ASK) and Phase Reversal-ASK (PR-ASK). ASK and PR-ASK Modulation are illustrated in Figure 4.

Amplitude shift keyed digital modulations are spectrally inefficient, requiring substantial RF bandwidth for a given data rate. Bandwidth efficiencies of 0.20 bits per Hertz of RF bandwidth are not uncommon for DSB-ASK. One approach to improving bandwidth efficiency is to use SSB-ASK. This is particularly important in European countries where bandwidth restrictions may preclude DSB-ASK.

The power efficiency of both DSB-ASK and SSB-ASK is dependent on the modulation index. With a modulation index of one or On and Off Keying (OOK) of the carrier, the lowest Carrier to Noise (C/N) required to achieve a given Bit Error Rate (BER) is obtained for DSB-ASK and SSB-ASK. Unfortunately, this also provides the least amount of RF power transport on the downlink to supply the tag with energy. Ideally, the off time of the carrier should be minimized so that the tag doesn't run out of power. The carrier to noise requirements should also be minimized to maximize ID read range. For many modulations these are conflicting goals.

PR-ASK is a modulation that can minimize the carrier to noise requirement in a narrowband while maximizing the power transport to the tag. This modulation has carrier to noise and bandwidth requirements more closely matching PSK than DSB-ASK, making it attractive for narrowband and longer-range applications. DSB-ASK is the least bandwidth efficient modulation, but the easiest to produce by On and Off Keying (OOK) of the carrier signal.

Data encoding considerations
Before modulation, the data must be encoded into a serial information stream. There are many types of bit encoding schemes available as shown in Figure 5, each with different strengths. Data encoding is critical for RFID applications due to such factors as the lack of precision timing sources on board the passive tag, challenging bandwidth requirements and the need for maximum RF power transport to energize the tag.

Manchester-L (Bi-Phase-L) and Pulse Interval Encoding (PIE) are popular for interrogator to tag (R=>T) communications. These coding schemes are based on transitions and are self-clocking, greatly reducing the complexity of the synchronization circuitry required in the power-starved tag.

PIE encoding is based on a given minimum pulse duration or interval such as 20 s. This period, called a Tari, is named after the ISO 18000-6 Type A Reference Interval. One and zero bits as well as special symbols like Start Of Frame (SOF) and End Of Frame (EOF) are composed of differing numbers of Tari periods. This makes the transmission length for a given number of bits variable, but since PIE encoding is self-clocking the variable length has little effect.

The Tari length is also the minimum pulse width for the modulated signal, an important factor in determining the bandwidth of the transmitted signal. The shorter the Tari length, the greater the bandwidth requirement for the signal. More recent standards such as the ISO 18000-6, Type C allow for several Tari lengths (6.25, 12 and 25 s) to accommodate differing worldwide regulatory spectral emission requirements.

Another important property for RFID Pulse Code Modulation (PCM) coding schemes is the DC spectral component. Backscattering tags modulate a carrier signal. The carrier signal is then filtered out as a baseband DC level back in the tag reader, leaving only the much weaker uplink modulation from the tag. Coding schemes in the tag require the uplink to the reader to have little or no DC energy to conflict with the carrier signal.

Miller and FM0 encoding share this property of little or no DC energy in their spectrums. ISO 18000-6 Type C further enhances the Miller encoding by offering different sub-carrier rates. One, two, four and eight times the sub-carrier frequency enable adjustment of the modulation encoding to optimize read range, speed or bandwidth.

Amplitude-based modulations used in many RFID systems are susceptible to rapid signal fading conditions. Pallets of tags traveling on forklifts past readers located between metal trucks and warehouses can undergo devastating multi-path conditions. Rapid Rayleigh fading or shadowing can be indistinguishable from amplitude modulation, leading to bit errors.

Another RFID consideration is some form of anti-collision protocol to enable reading all the tags in the interrogator's field of view. There are two basic types of anti-collision protocols, deterministic and probabilistic. Popular RFID protocols are the deterministic binary tree and the probabilistic ALOHA and slotted ALOHA approaches.

The binary tree method searches for tag IDs that fit a specific binary number while the probabilistic ALOHA protocol allows the tag to send its message and if the message doesn't get through, it simply tries again later until it does. The slotted ALOHA approach uses synchronization between all the tags, so communications packets are not interrupted mid-stream in the transmission. Additional efficiency gains are possible by using Listen Before Talk (LBT) schemes.

RFID testing overview
RFID systems, particularly those with backscattering passive tags, present a number of challenges for test and diagnostics. Timing measurements are of particular concern, as system readers can be required to read the ID data from many tags very quickly without error.

Most RFID systems use transient Time Division Duplexing (TDD) schemes, where the interrogator and tags take turns communicating on the same channel. To read many ID tags within a very short period of time with a serial TDD multiplexing scheme, standards call for precise timing on the data interchange, thus creating one of the more important RFID test challenges.