Voice over Internet Protocol, or VoIP, started out as a means of saving on long distance calls, but was considered too unreliable for business use, much less for the demanding healthcare environment.
Now, however, hospitals, physician practices, and clinics are eager to move voice calls onto their data networks to cut costs, improve communications and extend opportunities for patient care. So, what's changed?
Making voice calls more reliable over data networks has been achieved through the introduction of effective methodologies that overcome possible impairments that can interfere with voice quality — delay, jitter and packet loss.
Applications such as e-mail or file transfers are not affected by delay because the packets are delivered on a best-effort basis. But because conversations are expected to occur in real time, voice is very sensitive to delay. Too much delay interrupts the natural flow of the conversation. In a critical care setting, poor call quality would be unacceptable.
A network audit can reveal the need for more bandwidth to minimize delay, particularly on any lower-speed wide area network (WAN) connections that might be used between office locations. In such cases, congestion is most likely the cause of delay and is easily remedied with more bandwidth.
The routers at each location should also be examined. They may need to be reconfigured to give voice priority treatment on the network. For example, if voice and routine data packets arrive at the router together, but there is only enough free bandwidth to allow one onto the network, the data packets are held back while the voice packets go out first.
The routers' buffers may have to be upgraded to temporarily store the packets, ensuring their flow over the network. This step will usually ensure that delay stays within tight limits on your network so that voice conversations are never disrupted.
When some voice packets arrive with little delay followed by voice packets with greater delay, parts of the conversation on the receiving end will become uneven. This is called jitter, which is the variance in delay. When it comes to voice, it is better to have consistent delay rather than varying amounts of delay; the former is predictable and can be dealt with easier, while the latter is unpredictable and difficult to overcome.
The router buffers compensate for jitter by temporarily storing packets and controlling the speed at which they are offered to the network. But if a router's buffer is too large, it can cause delay as it waits to fill up with packets.
Packet loss is not as harmful to voice conversations as delay and jitter, but still can be disruptive if allowed to increase beyond tolerable limits. In routine data applications like e-mail and file transfers, lost packets are simply retransmitted before being put back in the right order for delivery to the proper recipient. Voice packets must be treated differently, since no time can be wasted waiting for their retransmission. As this would add too much delay to the conversation, lost voice packets are never retransmitted.
The amount of speech that is lost is usually so small that it is not noticed because the human brain is very adept at filling in the blanks. But if this problem were to become severe, the result would be "clipped" speech, which would disrupt the conversation. This could be remedied with more bandwidth to prevent packet collisions when the network becomes overloaded, or by equipping routers with the capability to rebuild lost packets by examining the surrounding packets that were received, thereby eliminating clicks or interruptions in the audio stream.
You should expect your IP phone system to be able to do everything a conventional phone system does, and more. Getting top performance from your IP phone system, however, will require an assessment of your network to determine if it is ready to handle voice traffic.
Jeff Nolte is president of Chesapeake Telephone Systems in Millersville, Md. (firstname.lastname@example.org)