Summary of Medical Device Wireless Coexistence and a Preliminary Plan for Work at the FDA

Abstract

As medical devices have increasingly incorporated wireless communication functionality the ability to function properly in the presence of interfering wireless signals has become of concern. In the MICS and Med Radio bands the problems are compounded by the fact that medical devices use the bands on a secondary basis. The WMTS bands were created in response to TV station interference on the MICS and Med Radio bands. Medical devices are primary users of the WMTS bands, but interference from TV stations is still possible and only 14 MHz of bandwidth is available. However, in May of 2012 the FCC set aside additional spectrum for use by medical devices in the range of 2360 – 2400 MHz. Medical devices in this band are secondary to aeronautical mobile telemetry and radio astronomy users, but given the 40 MHz bandwidth and that devices won’t have to compete for spectrum in the 2.4 GHz ISM band it will likely be heavily used.

There are two tracks of work on which the FDA could begin coexistence testing, both of which involve developing a generally applicable coexistence test method. One starting point would be to investigate the feasibility of testing coexistence using the National Instruments 5663E vector signal generator to create interference. Typically, coexistence testing is performed using two or more test boards to generate interference. The boards are set up to send data back and forth using a common wireless protocol. A medical device can then be tested by exposing it to the interference while it communicates with a second device. Parameters such as packet loss or bit error rate can be monitored. However, it would simplify coexistence testing if the NI signal generator can be programmed to generate interference that is similar to an actual network of communicating devices. The NI 5663e vector signal generator can be programmed to transmit with random timing mimicking the transmissions in a real network. [MAE1]

Comparing coexistence studies performed by different groups at different locations is difficult due to variation in factors such as received signal strength and fading. A second track of work is to investigate the feasibility of testing LOS and NLOS configurations in an anechoic chamber. It is difficult to compare the results of coexistence studies performed by different groups at different locations due to variation in factors such as received signal strength and fading. The WECAD group at the University of Oklahoma has developed NLOS and LOS testing protocols that it hopes will make reproducible coexistence measurements possible in different environments. A second track of work then is to investigate the feasibility of testing LOS and NLOS configurations in an anechoic chamber. Work at the FDA could be focused on testing this protocol to determine the reproducibility of results obtained for different test configurations in the 10 meter chamber.

Introduction

As medical devices have increasingly incorporated wireless communication functionality the ability to function properly in the presence of interfering wireless signals has become of concern. These concerns were heightened in 1998 when a Dallas TV station tested a digital TV transmitter which interfered with transmissions of cardiac telemetry systems being used at two nearby hospitals. In response to this incident, the FCC created the WMTS bands in which medical devices are primary users. However, TV station interference is still possible since the transmitters often emit energy outside of the TV bands. In the MICS and Med Radio bands coexistence problems are compounded by the fact that medical devices use the bands on a secondary basis.

However, in May of 2012 the FCC set aside additional spectrum for use by medical devices in the range of 2360 – 2400 MHz. The lower 30 MHz, 2360 – 2390 MHz, are for use only in health care facilities while the upper 10 MHz can be used anywhere. Medical devices in this band are secondary to aeronautical mobile telemetry and radio astronomy users, but given the 40 MHz bandwidth and that devices won’t have to compete for spectrum in the 2.4 GHz ISM band it will likely be heavily used. [MAE2]

Due to problems with interference and limited bandwidth in the MICS, MedRadio and WMTS bands many medical device manufacturers have been incorporating wireless technology that utilizes the 2.4 GHz ISM band instead. [MAE3]In this frequency range devices can encounter many interferers including WiFi networks, microwave ovens, baby monitors, and cordless phones, in addition to devices operating with short range wireless protocols like Bluetooth and ZigBee. Devices employing these latter two protocols include headsets, wireless mice, keyboards, mobile phones, and medical sensors. A partial list of medical devices currently incorporating wireless capability include respiration and cardiac sensors, blood pressure sensors, EKG sensors, oxygen saturation sensors, glucometers, and ICDs. As the proliferation of wireless gadgets increases, congestion in the 2.4 GHz ISM band grows thereby making coexistence more problematic. It becomes difficult to guarantee that the critical communications of a medical device operating in this band can be sent and received properly. [MAE4]

Currently, no guidance for assuring adequate wireless communication performance of medical devices exists. A standard test methodology for medical devices using wireless capability in the ISM band would be beneficial in assuring coexistence of these devices in a clinical setting. Such a standard should be easily extended to cover the new MBAN band. The protocol should evaluate the medical device under test for potential interference in not only the physical (PHY) layer, but also the medium access control (MAC) layer. It is not enough to expose medical devices with interfering signals using in-band or co-channel frequency signals and measure electromagnetic compatibility.

PHY layer interference occurs when packets collide at a receiver possibly resulting in lost or garbled packets. To lower the probability of PHY layer interference, many wireless protocols implement a clear channel assessment (CCA) function in the MAC layer. CCA is used to ensure that a channel is free by sensing the medium before allowing a transmission. MAC layer interference manifests itself as lost packets when the channel is repeatedly sensed busy causing transmissions to be deferred until buffer overflow occurs.

There is little published work that deals with coexistence in the medical device bands as compared to the 2.4 GHz ISM band. In the 2.4 GHz range most of the papers are not specific to medical systems, however, since the papers cover protocols that are used in medical devices they can still be considered relevant. Kamerman and Erkocevic published a paper in 1997 examining interference of microwave ovens to WLAN networks. Over the following several years a number of papers examined the coexistence of 802.11b with Bluetooth. [MAE5]The papers typically report results of an analytical treatment of coexistence parameters, simulations, experimental investigations or some combination of these. Researchers have studied many different aspects of the coexistence problem including power, packet duration, data transfer rate, modulation, frequency hop rate, number of nodes, spectrum utilization, signal to interference plus noise ratio (SINR), pattern of traffic, etc.

In 2003 IEEE wrote a recommended practices document (802.15.2 [7]) reporting the development of simulation and analytical models addressing coexistence problems between 802.15.1 and 802.11b. Soon after, coexistence work focused on 802.15.4. The findings can be categorized into those looking at PHY layer problems and those focusing on MAC layer issues. PHY layer work can be subdivided into the three parameters frequency, space, and time. For the MAC layer the literature focuses on CCA and the size of the 802.15.4 packets.

There are also papers that have specifically investigated medical wireless applications, many of which focus on medical body area networks (MBANs) [2-4]. The papers generally conclude that achieving packet error rates below 10% with an adaptive frequency hopping protocol will require that spectrum occupancies be kept low (below 20% according to [3]) and the number of channels must be greater than 20 [2]. Due to the overhead and possibility of collision associated with CSMA and the limited ability of AFH to provide acceptable performance the IEEE 802.15 Task Group 6, which is working on a standard for body area networks, has put forth a time division multiple access (TDMA) scheme. TDMA protocols require synchronization between all nodes since time is divided into slots. This method eliminates collisions and can maximize the channel efficiency, but is easiest to implement when the network topology is static. The static topology requirement is often met by MBANs.

Since most coexistence work has focused on the 2.4 GHz ISM band and many researchers will likely begin to focus more on the new 2.36 GHz medical band it is reasonable for the FDA to begin work by investigating coexistence in the 2.4 GHz band. This is also where Dr. Refai's group from the University of Oklahoma is concentrating its efforts.

Experimental Methodology

A wide variety of coexistence test configurations are found in the literature which makes it difficult to show a “typical” configuration. However, fFigure 1 shows one method for testing coexistence. In this setup two competing networks are created using test boards or actual devices (one shown in blue and the other shown in green). Test boards have the advantage of providing more control over the network protocol parameters. Two The two blueof the test boards represent a pair of medical devices that send data back and forth which could be things like. These might be an EKG sensor that sends data to an EKG monitor using ZigBee (802.15.4), for example. The other green pair of test boards represents a WiFi network using the 802.11g protocol. There are no standard methodologies specifying the type or placement of the test boards or the nature of the physical environment in which the tests are performed even though the testing results are strongly affected by parameters like received signal strength, network utilization, line-of-sight vs. non-line-of-sight node placement, etc.

Figure 1: A coexistence test setup with 4 nodes. Blue nodes communicate with each other in the presence of an ongoing communication between the green 802.11 nodes.

Whatever configuration is used theAfter configuring the devices in a specific layout they are tested for coexistence by sending traffic over both networks simultaneously. By monitoring parameters like packet loss or bit error rate allows for the degree of interference can to be measured quantitatively. Variations in these parameters can be investigated due to changes in packet length, network utilization, received signal strength, CCA threshold, distance between nodes, etc.

Given the lack of standardization in coexistence testing, the FDA’s work would best be focused on developing standardized methods for testing, as currently, it is difficult to compare coexistence test results performed in different labs.

FDA activities

There are two tracks of work on which the FDA could begin coexistence testing. One starting point would be to investigate the feasibility of testing coexistence using the National Instruments (NI) 5663E vector signal generator (VSG) to create interference. This would eliminate the necessity of using test boards and would make testing quicker and easier. Dr. Refai’s group has found that using a typical signal generator that outputs a signal with uniform timing on a single frequency channel causes a competing medical device to behave differently as compared to when the medical device is exposed to an actual network. However, the NI VSG can be programmed to transmit with random timing. It gives full control of the interference generation parameters, so that any wireless protocol can be emulated.

One complication with using a VSG to create interference is that a VSG cannot sense the wireless channel and, thus, can’t assure that its transmissions won’t collide with those from other devices. However, the lack of sensing may not be detrimental as long as the VSG is made to generate interference possessing the same statistical signature as is present in a real clinical environment.

The second track of work is to investigate the effect of the physical environment in which medical devices are used. Currently, it is difficult to compare the results of coexistence studies performed by different groups at different locations due to variation in factors such as received signal strength, fading, node placement, etc. These parameters are strongly affected by the physical placement of the devices, whether line-of-sight (LOS) or non-line-of-sight (NLOS), and the surrounding environment. It is reasonable to investigate whether medical devices that may be used in a NLOS configuration must be tested in NLOS configurations or whether LOS testing is sufficient. Conti, et al., have found differences in packet error probabilities for LOS and NLOS conditions due to differing statistics that govern the signal to noise ratio [1]. The WECAD group at the University of Oklahoma has developed NLOS and LOS testing protocols that it hopes will make reproducible coexistence measurements possible in different environments. Work at the FDA can be focused on testing these protocols to determine the reproducibility of results obtained for different test configurations in the 10 meter chamber.

References

[1] A. Conti, et al., “Bluetooth and IEEE 802.11b Coexistence Analytical Performance Evaluation in Fading Channels,” IEEE J. Selected Areas in Comm., v21 no. 2, Feb. 2003, pp 259 – 269.

[2] D. Davenport, B. Deb, F. Ross, “Wireless Propagation and Coexistence of Medical Body Sensor Networks for Ambulatory Patient Monitoring,” Body Sensor Networks (BSN 2009), Berkeley, CA, June, 2009.

[3] H. Hellbruck and T. Esemann, “Limitations of Frequency Hopping in 2.4 GHz

[4] Gengfa Fang; Dutkiewicz, E.; Huq, M.A.; Vesilo, R.; Yihuai Yang; , "Medical Body Area Networks: Opportunities, challenges and practices," Communications and Information Technologies (ISCIT), 2011 11th International Symposium on , vol., no., pp.562-567, 12-14 Oct. 2011

[5] Huasong Cao; Leung, V.; Chow, C.; Chan, H.; , "Enabling technologies for wireless body area networks: A survey and outlook," Communications Magazine, IEEE , vol.47, no.12, pp.84-93, Dec. 2009.

[6] “Limitations of Frequency Hopping in 2.4GHz ISM-Band for Medical Applications due to Interference,” 1st IEEE International Workshop on Consumer eHealth Platforms, Services and Applications, pp. 242 – 246, 2011.

[7] IEEE Recommended Practice for Information Technology—Telecommunications and Information Exchange Between Systems—Local and Metropolitan Area Networks—Specific Requirements Part 15.2: Coexistence of Wireless Personal Area Networks With Other Wireless

Devices Operating in Unlicensed Frequency Bands, IEEE Standard 802.15.2-2003.

[MAE1]This seems rather redundant and could be combined into one statement.

[MAE2]Depending on the audience that is reading this, 40MHz may seem like a lot or a little bit of bandwidth. Maybe you could throw one statistic in here on bandwidth of common devices now OR suggest how the addition of 40MHz will help the medical industry. It’ll just give a nice scale perspective to your audience who may not be as familiar with the subject.

[MAE3]Sentence does not make sense. I think you were trying to say that medical devices in the ISM band are ban?

[MAE4]I don’t think you need this sentence because the last sentence talking about the problem of coexistence says this!

[MAE5]Reference the papers