Originally Published MDDI September 2004

Integrated Circuits



Understanding radio protocols and requirements is the first step in successfully using wireless communication in medical applications.

Carl Falcon

A wide variety of medical applications, such as implantable devices and telemetry equipment, are increasingly using wireless communication. And at the heart of this trend is the need for low-power radio application-specific integrated circuits (ASICs). In order to select the appropriate radio ASIC, it is critical to first understand the protocols and requirements needed for the particular medical device. This article provides an overview of the protocols commonly used in medical radio applications. It also reviews the essential factors to consider when selecting a frequency band, topology, and transmission protocol.

Common Medical Radio Requirements

Medical radios are usually driven by two primary requirements. First, medical radios must consume very little power so that they can last on battery power for months or years. Second, medical radios must often be added to other low-cost, small-sized components such as sensors. This consideration requires that these radios be low cost, have a simple design, and need few external components.

Medical applications are usually indoors, which typically translates into a relatively short range over which data must be transmitted. In addition, the radios need to transmit only a small amount of data, and those data are transmitted only infrequently. 

A short data range means that the radio design can be kept simple and that the transmitter power can be kept low. The infrequent data transfer means that medical radios can sleep a large portion of the time. Both the short data range and infrequent data transfer of these radios help to minimize power requirements. 

Of course, the requirements of a medical wireless system go beyond low power and low cost. Table I lists additional requirements that should be considered when reviewing possible wireless system architectures.

Selecting a Frequency Band

Several licensed frequency bands exist specifically for medical applications. In addition, many low-cost applications operate in the unlicensed frequency bands known as the industrial, scientific, and medical (ISM) bands. No license is necessary to operate a device in this band. 

Most radio ICs are transceivers, which means the IC can both transmit and receive data. A transceiver is often paired with a simple microprocessor or other support circuitry that directs the baseband operations, such as controlling the flow of information between the real-world interface and the transceiver.

In the simple wireless system shown in Figure 1, the analog data from a sensor measurement is first converted from its continuous-time analog signal into a discrete-time digital signal. A baseband processor then adds some error-detection information and formats the data for transmission. The baseband data are sent to the transceiver, encoded onto a radio-frequency (RF) signal and broadcast by the transmitter, detected by the receiver, and finally decoded back into digital data. Low-power transceiver ICs usually operate between 300 MHz and 1 GHz. The frequency spectrum between these limits contains many bands where both licensed and unlicensed medical and industrial equipment can operate (see Figure 2). Table II details the common North American and European frequency bands that are suitable for certain medical applications.

In the United States, the Federal Communications Commission (FCC) has set aside a few frequency bands for medical data transmissions only. In 1999, FCC set aside the medical implant communications service (MICS) frequency band. The MICS band, located in the frequency range of 402–405 MHz, is reserved specifically for wireless data communications between implanted medical devices and external equipment. 

Wireless medical telemetry enables monitoring equipment to remotely and unobtrusively observe several patients at one time. Such telemetry systems transmit real-time physiologic data, so it is critical to ensure that data are not lost or delayed. More and more radios for nonmedical applications are operating in the ISM bands, 

increasing the likelihood of signal loss and interference. FCC, therefore, has set aside frequency bands specifically for wireless medical telemetry services (WMTS). The radio spectrum designated for WMTS includes the 6 MHz between 608 and 614 MHz.

Topology

Wireless receivers and transmitters are often arranged in a star configuration, in which a centrally located transceiver communicates with one remote location (a point-to-point topology) or with several remote locations (a point-to-multipoint topology) simultaneously. In a point-to-point architecture, only one transmitter and receiver pair is communicating at any given time on a specific carrier frequency. The central node takes the role of a master coordinator, while the remote location is a subordinate. Point-to-point communications can be simplex (one way only), half duplex (first one direction and then the other, sequentially), and full duplex (simultaneous communication in both directions).

Networks that are more complex may have multiple transmitters and receivers that can communicate with one another as peers. Within a network, one of the central nodes is designated as a coordinator. The coordinator is tasked with waking up other subordinate devices on the network out of a low-current sleep mode just before data are to be transmitted. Coordinator transceivers can also talk to one another as peers.

Mesh networks allow wireless devices to talk indirectly to one another even when the two devices cannot see each other. A transmitting device can pass data to its neighbor, which in turn can pass data onto its next neighbor, and so on. In this way, a mesh network can be used over a far-flung network, say, between the top and bottom floor in a high-rise building. Figure 3 illustrates two different network topologies.

Selecting a Transmission Protocol

Transmission protocols define the way data are to be encoded, transmitted, received, and decoded. Protocols also define any error detection methods as well as techniques to minimize interference and distortion. Protocols are usually optimized for the given application to reduce the overhead associated with error correction and thus result in lower-cost and lower-power operation.

Medical equipment manufacturers concerned about the interoperability among radios from different manufacturers should consider using an open-standard protocol such as Bluetooth or ZigBee. Figure 4 maps several open standards against the unlicensed frequency bands. Open-standard protocols can also allow for a complex networking of transmitters and receivers such as those operating in a mesh-type network. 

However, standard protocols give up some flexibility, and considerable overhead is required to ensure compatibility and interoperability. Bluetooth and ZigBee radios are also designed for higher-frequency bands, such as the 2.4 GHz ISM band. The wireless local-area-network (WLAN) and wireless personal-area-network (WPAN) open standards were developed to address high-speed data transmission requirements. Because medical applications typically require only low data rates, the WLAN and WPAN standards (e.g., IEEE 802.11a, b, or g) are rarely appropriate for small implantable medical devices. 

In some cases, proprietary protocols can fit a given medical application better than a standard. Often, proprietary protocols are the only ones available for MICS band operation. A proprietary protocol can offer more flexibility and provide lower-power operation if interoperability is not a concern. A proprietary protocol can also be less expensive overall.

A few considerations should be taken into account when choosing a protocol and an associated frequency band. Among these considerations are:

•Geographic location. The medical frequency bands used worldwide do not necessarily coincide from country to country (see Table II). The 2.4 GHz band is generally available in most locations, but it may be restricted depending on the country. Other, lower-frequency bands are restricted to a geographic location.
•Interoperability. If the application requires that devices from several manufacturers must work together, then a specific communications standard must be used. That standard will define the frequency range. The newly adopted IEEE 802.15.4 standard, commonly known as ZigBee, addresses the 868 MHz, 902–928 MHz, and 2.4 GHz ISM bands and ensures interoperability between various radio devices. 
•Interference. Multiple wireless standards often coexist in unlicensed frequency bands. For example, 900 MHz cordless phones operate in the 902–928 MHz band. In addition, the 2.4 GHz band is usable in most locations, making it a popular choice for a variety of wireless communications protocols. Significant microwave oven operation also can plague that band.

Low-data-rate radios can usually host both the protocol and application control in a simple 8-bit microcontroller. Typical applications use only between 4 and 30 Kbyte of code and less than 300–4000 bytes of RAM, depending on whether the radio is a full-function controller or a reduced-function device.

By comparison, more-sophisticated radios, such as Bluetooth, require a dedicated core to manage the baseband protocol stack. They also need a separate host device to manage the logical link control and adaptation protocol, protocol interfaces, and applications. In order to achieve this, between 100 and 200 Kbyte of code and about 150 Kbyte of ROM may be required for a full software stack.

A clear understanding of the trade-offs between open standards and proprietary protocols can be developed by comparing two similar radio transceivers that operate in the same 868 MHz (EU) and 902 MHz (U.S.) unlicensed medical bands. Table III shows how two different transceiver designs operating in the same frequency bands address the wireless system requirements noted in Table I. Transceiver A follows a more-complex open-standard protocol (IEEE 802.15.4), while Transceiver B follows a simpler but proprietary protocol. 

Environmental Requirements

In a medical environment, data transmissions through a wire or along a circuit board trace are always vulnerable to interfering signals, coupled noise, electromagnetic interference, and noisy or shifting power and ground planes. 

A common method of overcoming transmission problems is to use the digital data to modulate a much higher frequency signal, called a carrier signal. Carrier modulation techniques convert the digital signals into frequencies that are inherently more immune to noise. 

Because the shifting carrier signal is an ac signal with no ground reference, the carrier signal is not affected by unstable or noisy ground planes. The carrier frequencies can be tuned specifically to avoid electrical noise, and the receiver circuits can be designed to reject out-of-band noise. The receiver can also be designed to reliably detect very small changes in frequency that can yield long data ranges with very little power.

Carrier Modulation Techniques

A number of techniques can encode digital data by varying either the amplitude, the frequency, or the phase of a carrier signal. Examples of each technique are shown in Figure 5. These modulation techniques include: 

•Frequency shift key (FSK) modulation, which uses two different carrier frequencies to represent the logic high and logic low of digital data.
•Amplitude shift key (ASK) modulation, which uses one frequency but varies the amplitude of the RF carrier to represent the logic high and logic low of digital data. A variation of ASK is called on-off shift key (OOSK) modulation, where one of the two amplitudes is zero. The OOSK signal is a series of alternating-frequency bursts and quiet periods representing the logic high and logic low of digital data. ASK is more susceptible to noise than FSK, but it is often easier to implement and detect. 
•Phase shift key (PSK) modulation, which encodes the digital data by altering the phase of the carrier frequency. Biphase shift key (BPSK) modulation shifts the phase of the RF carrier by 180 in accordance with a digital bit stream. The receiver performs a differential coherent detection process whereby the phase of each bit is compared with that of the preceding bit. 

Table III also shows that Transmitter A uses the biphase modulation technique, while Transmitter B uses an on-off amplitude modulation technique. BPSK is more difficult to implement but can offer a 6-dB advantage in signal-to-noise ratio over OOSK for a given carrier level.

Multipath Propagation Problems

Once the digital data have been encoded as an RF carrier signal, environmental obstacles can affect the data signal path.

Multipath propagation occurs when an RF signal takes different paths when propagating from a transmitter to a receiver. While the signal is en route, obstacles such as walls, chairs, desks, and other items get in the way and cause the signal to bounce in different directions. A portion of the signal may go directly to the destination, and another part may bounce from a chair to the ceiling, and then to the destination, as illustrated in Figure 6. Therefore, some of the signal encounters a delay, increasing the length of path to the receiver. 

Multipath delay causes the information symbols represented in a radio signal to overlap, which confuses the receiver. This is often referred to as intersymbol interference (ISI). Because the shape of the signal conveys the information being transmitted, the receiver will make mistakes when demodulating the signal's information. If the delays are great enough, bit errors occur. The receiver won't be able to distinguish the symbols and interpret the corresponding bits correctly. 

Radio Interference Problems

The proliferation of many types of common wireless devices in the unlicensed ISM bands can corrupt the data signals of medical radios operating in the same band. Airborne radiolocation systems also share these bands.

When these unwanted RF signals transmit at the same carrier frequency as a medical radio signal, they can cause an interruption or corruption of that signal. Corrupted signals arrive with errors, missing bits, or not at all, and the receiver cannot recover the data. Depending on the sophistication of the receiver-transmitter set, the transmitter may be asked to resend the data, which in turn adds overhead to the network and causes delays.

Multipath and Interference Reduction 

Depending on the application, radio interference may be a minimal concern if the environment is relatively quiet. However, indoor short-range radios must be able to account for multipath problems, especially in the ISM bands.

Both active and passive techniques are used to account for multipath fading. Passive techniques generally cost much less. These techniques rely on diversity antenna systems—two antennas for each radio—which increases the odds of receiving an uncorrupted signal.

Diversity antennas are physically separated from both the radio and from each other to ensure that one of the antennas will encounter fewer multipath propagation effects than the other antenna (see Figure 7). The receiver uses two separate receiver paths to filter the signal and then decides which signal is the best choice for demodulation. 

Active techniques employ spread-spectrum technology to address multipath fading and RF interference. Two commonly used techniques are direct-sequence spread spectrum (DSSS) and frequency-hopping spread spectrum (FHSS). DSSS mixes a data signal with a constantly changing pseudorandom noise signal that spreads a continuous data signal over a wide range of frequencies. This leaves enough room for lower-frequency elements of the DSSS signal to reflect off obstacles much differently than the higher-frequency elements of the signal.

FHSS transmits an intermittent data signal for a short period at one frequency and then hops to other frequencies in a pseudorandom pattern. FHSS avoids interference from other radios by hopping on narrow channels over a wide range of frequencies; transmitting on strong, clear channels; and avoiding noisy or faded ones. Frequency hopping is generally regarded as having better noise immunity than DSSS; however, the data rate is lower. 

Data Integrity Checks

Error detection and correction are also important in mitigating the effects of interference and multipath fading. The receiver can check for the presence of errors through an error detection process. Basic designs use a simple checksum, but a cyclic redundancy code (CRC) check is necessary for more-reliable data transmission. With the latter process, the transmitting transceiver adds a CRC to every data packet that is sent. A data packet containing an error will not compute correctly. The drawback of added overhead is more than offset by ensuring the received data are accurate. For low-data-rate radios, a CRC of 16 bits is adequate; for high-data-rate radios, a CRC of 24 bits is ideal.

In a point-to-point topology, a simple handshake protocol between the receiver and transmitter known as automatic retransmit request (ARQ) requires the receiver to send an acknowledgment if data have been received with no errors. If an acknowledgment is not received by the transmitter, the transmitter resends the data.

With complex network topologies, such as point-to-multipoint or peer-to-peer topologies, the protocol must enable one receiver to accommodate multiple transmitters broadcasting at the same time. The primary scheme is called carrier sense multiple access with collision detection (CSMA/CD). CSMA/CD is a set of standard rules that determine how network devices should respond when two devices attempt to use the same channel. The handshaking process between receiver and transmitter is similar to ARQ. After detecting a collision, a transmitter waits a random delay time and then attempts to retransmit the data. If the transmitter again detects a collision, the device waits twice as long to try to retransmit the message. This is known as exponential backoff.

Because of data retransmissions, the data rate throughput is slow when multipath propagation is significant. The reduction in throughput depends on the environment. Radio signals in homes and offices may encounter 50 nanoseconds of multipath delay. Metal machinery and racks in a plant, however, provide a lot of reflective surfaces for RF signals to bounce from and take erratic paths. Signal delay in a manufacturing plant could be as high as 300 nanoseconds. It is essential to be aware of multipath problems in warehouses, processing plants, and other areas full of irregular, metal obstacles.

Conclusion

Understanding the protocols and requirements needed for the particular medical application is a crucial step in selecting the appropriate low-power radio ASIC. The protocols and requirements should be reviewed and considered carefully to ensure that the proper frequency band, topology, and transmission protocol are used for a device. A thorough understanding of the environment is also critical to identifying and implementing modulation techniques to reduce multipath propagation, noise, and interference, and to ensure data integrity.

Bibliography

Code of Federal Regulations. Title 47 Telecommunication, Part 15. Washington, DC: National Archives and Records Administration, August 20, 2002.

ETSI EN 300 220-1, “European Standard (Telecommunications series),” V1.3.1, September 2000.

ETSI EN 300 328, “Candidate Harmonized European Standard (Telecommunications series),” V1.4.1, November 2002.

ETSI EN 301 839-2, “Candidate Harmonized European Standard (Telecommunications series),” V1.1.1, April 2002.

Falcon, Carl. “Adding a Low Data Rate Radio ASSP to an ISM Application.” Microwave Journal, October 2003.

Sklar, B. Digital Communications, Fundamentals and Applications. Englewood Cliffs, NJ: Prentice Hall, 1998. 

Carl Falcon is a strategic marketing manager for AMI Semiconductor (Pocatello, ID). 

Copyright ©2004 Medical Device & Diagnostic Industry