This was an interesting week on the Biomed Listserv. One list member kicked things off with the following observation:

I have heard sales reps for certain new IV/PCA pumps claim that if we buy their new pump with 802.11 they can be located on computer by which AP they are closest to. I haven't seen a demo yet but it sounds interesting.

This is a capability offered by most of the current 802.11 infrastructure vendors. That got the following reply describing someone else's experience...

Doesn't work as good as they say.

I have seen a pump connect to an access point through a window, across a courtyard, and on another floor. Also, the pumps I have seen recently "hold" on to an access point till the signal is unusable, then it looks for another. The pump can be sitting under access point "C", and be "connected" to access point "A". The transmitter/receiver boards work like cell phones, they ramp up the power, this is why two of the manufacturers I have looked at are having a hard time with battery life. Using this method is still better than walking around for two days with a clip board looking for equipment.

The positives of the 802.11 system [for RFID] far outweigh the few negatives... in my opinion. It just isn't perfected yet, and when it is perfected, there will be something new that's bigger and better than 802.11 out there.

The thing that struck me in this exchange was not the use of 802.11 for an RFID system, but the comments about WLAN performance. (I've posted before on the advantages and disadvantages of 802.11 based RFID systems.) This thread hit on one of the dirty little secrets of wireless networking - and one that some (maybe most) medical device vendors and hospitals don't quite appreciate yet. Then there was this:

The two biggest problems with wireless
LANs are 1) 99% of wireless devices are designed with the technically
inclined consumer in mind, not enterprise users and certainly not
healthcare enterprise users, and 2) there is still a great dearth of
individuals who don't understand RF propagation characteristics well
enough to properly design and install a robust wireless LAN capable of
scalable capacity. (Most commonly seen mistake: not understanding a
fancy RF survey software package is only a tool, not a complete
solution to the puzzle of WLAN [wireless local area network] design.)

I'll forgo the details,
but it seems incredible to me the number of people I've talked with who
tout they have a showcase wireless system, yet when you start asking
for details they outline several crippling limitations that are
obviously due to a complete lack of understanding of how these systems
really work. In my opinion, there are far too many people (both device
manufacturers and hospitals) believing the hype and throwing money at
problems that don't exist in order to keep up with the guy across the
street. The one good thing I can say is that folks are beginning to
realize there is more to it than throwing up a bunch of access points
and sticking wireless NICs in the back of devices.

There are a number of key performance capabilities or metrics that are critical to wireless medical device performance. Roaming across subnets, which could happen when you take a patient from a nursing unit to radiology, is a function of infrastructure - not all access points/switches can support roaming. The capacity of a WLAN to support a growing number of wireless devices in the same area is a function of how the network's designed and whether access points can support multiple channels and VLANs at the same time. Currently, some medical device vendors design WLANs with redundant access point coverage.

Wireless medical device vendors have to validate their products' performance on WLANs. The extent of validation required depends on the nature of the data. Continuous patient monitoring, or any device that can generate life threatening alarms, must be extensively tested to ensure safe performance. Simple data capture for paperless charting, like spot vital signs, requires minimal validation. The easy way out for medical device vendors is to specify a separate WLAN using their preferred hardware vendor, equipment and design - this way they can validate against a set infrastructure and keep things simple and straightforward. A separate network also makes it quick and easy to trouble shoot because the variables found in a general purpose WLAN are eliminated.

Sadly, market requirements (and trends) are moving in the opposite direction towards running medical devices on a general purpose WLAN. As more medical devices gain wireless capabilities, hospitals are not going to have the luxury of putting up dedicated WLAN infrastructure for each one. Proof of this trend can be found in the initial success of companies like InnerWireless and MobileAccess, and product strategies like Draeger's OneNet. It's also nice that Cisco has a nice new up to date WLAN product offering, thanks to the acquisition of Airespace. Any wireless device now in development that is based on a private network is going to face some unanticipated rework.

But the real challenge for medical device vendors is not the infrastructure, it is the embedded radio. Even the most sophisticated and feature rich infrastructure cannot anticipate all the needs of the different wireless clients connected to it. Much of the wireless reliability and performance of a medical device is dependant on the wireless client in the medical device. Just establishing and maintaining patient context in a wireless device that goes in and out of coverage and moves between access points and subnets is a challenge. To preserve battery life sophisticated power management must be employed. In continuous monitoring applications, great care must be taken to optimize RF performance and ensure that no data is lost. Continuous systems are also sensitive to latency, so that must be minimized as well. And finally, any wireless device must be able connect to the hospital's network - which requires certain flexibility in WLAN radio configuration.

Most hospitals have deployed 802.11b networks, while newer networks support 802.11b/g; some hospitals have deployed infrastructure that supports all three current standards: 802.11a/b/g. Most medical devices using 802.11 use "b" radios. Few medical device applications require the bandwidth of 80.211a or 802.11g - a typical multi parameter patient monitor only consumes about 12kbs bandwidth. As 2.4GHz gets more crowded interest in 802.11a will probably grow (upper right is a shot of an 802.11a radio chip). There a number of relatively new standards that can facilitate network integration. These new standards include 802.1x and 802.11i for security, 802.11e for quality of service (QoS) - that helps ensure that life critical data gets through - and 802.11h for improved transmit power control.

There are a lot of technical requirements that go into supporting truly mobile devices, especially in life-critical applications. Whether you're selecting a WLAN card for your Draeger patient monitors (they support your choice of Wi-Fi certified WLAN cards) or evaluating device vendors with embedded 802.11 radios, it is important to ask the right questions and thoroughly test in your environment.