“Star Trek: The Next Generation,” characters in the 24th century communicated by tapping a badge and speaking (“Picard to Engineering”). Magically, the badge would contact the person no matter where they were on the ship. Here in the 21st century, Vocera Communications Inc. has introduced a wireless LAN-based interactive voice response system that uses lightweight badges similar to the “Star Trek” system.
The Vocera Badge connects via an installed wireless LAN to a Vocera server and is operated largely by voice command. (As an option, you also can connect to analog phone lines by connecting Vocera to a PBX, Centrex or other system.) With a bit of work, these no-hands communicators can become very useful where wireless LANs based on 802.11b deployments are working.
Vocera doesn’t control the most critical component of the system – the 802.11b infrastructure. We found that if a wireless LAN can be “killed” or has “bare-spot reception,” so can Vocera. But when configured correctly, Vocera can operate during very high wireless LAN network loads without losing quality or functionality. It won’t replace a PBX or your business phone system, but this can augment and “cut the cables” for mobile employees. Companies that use walkie-talkies might find the Vocera system sleeker, more feature-filled and less cumbersome.
The system
Vocera shipped us a preconfigured Dell Computer Corp. 4500 server with Windows 2000 Advanced Server installed. Normally we prefer testing the components that end users/deployment personnel would use, but Vocera ships 100 percent of its server systems through value-added resellers, which are required to deploy a ready-to-run/configured system. So we let Vocera ship its server almost completely configured. However, some work on the server was required.
The Vocera server includes a base platform (the Dell 4500 in our case), Win 2000 Advanced Server (Win 2000 Professional can be used and we recommend that), Vocera software, Nuance voice-recognition software and an optional multiport Intel/Dialogic phone board. The total Vocera software installation took about 10 minutes.
The badges (we received four) are based on 802.11b (other 802.11 variants are not available) and initially are programmed via a configuration download through an access point. The access point requires a specific IP address (in a specific range) to download the configuration. Installation is more difficult than it needs to be. Whether the choice is static or dynamic IP address, each badge must have an initial wireless LAN adaptation setup procedure. This ritual is oddly cumbersome as all configuration must come from the wireless LAN – there is no equivalent to a cell phone “base station-to-PC” connection for initial configuration.
After some additional minor glitches, we configured the badges and set up the group/user database. Vocera cannot read an Active Directory or otherwise use Lightweight Directory Access Protocol or another directory service to import user or group information into its database. Vocera uses the MySQL database and Apache/Tomcat Web server (which requires shutting down or moving Internet Information Server’s Web service ports if they were installed).
A genie in the badge
Next was configuring how the voice recognition (humanized in voice responses through the badge) would sound. Male or female voices are offered – we chose the male voice, but in retrospect the female voice was easier to understand.
The Genie application is analogous to an older voice-recognition-driven application called the Wildfire Assistant from Wildfire Communications Inc. When we used a badge for the first time, we pressed the call button on the badge, and the male voice asked for our name. We spoke our name, and the badge repeated the name back to us – when we responded, the badge became registered to us.
Pressing the call button let us locate individuals (and groups) and bring them into the conversation. Controls on the side of the badge are used to check messages that can be stored during off-time, although Vocera is not a standard voice mail server. There is a Do Not Disturb mode that users can select – voice mail also kicks in when users are out of range or offline.
The optional voice board, a Dialogic D120JCT-LS 12-port board, was connected to two analog trunks (the system can support two Dialogic boards for a total of 24 concurrent ports). We successfully used Vocera to dial internal and outside extensions. Unfortunately, the software doesn’t recognize a command such as “Call George” and numbers need to be spoken. Vocera does remember dialing prefixes for outside lines and long-distance prefixes.
Testing the badges
We used several tests to discern the voice-recognition quality of the Vocera/Nuance combination. High ambient background noise (such as 82db/A average from equipment cooling fans in our network operations center) made voice recognition difficult; we were forced to repeat our selections several times before they were recognized. In low background-noise areas (ambient noise under 72db/A average), recognition was nearly flawless.
When the badge hangs from a lanyard that is worn around the neck, the speaker faces away from the user. Additionally, the microphone embedded in the badge picks up lots of ambient sounds around the user. Using the earphone/microphone combination in conjunction with the badge helped the system recognize voice immensely and also kept privacy high. Without this earphone/microphone combination, communications via a badge are fairly public.
We also subjected the system to a variety of background Ethernet and wireless LAN data traffic to see how it affected communications. On a quiet network, Vocera subjects conversations to about a 1-second delay between speaker and listener. We subjected all our access points to a large amount of traffic using either a 4G-byte FTP file transfer or with our Spirent Communications traffic generator, and could cause outages only when traffic rose to denial-of-service levels – effectively jamming all the access points in our wireless LAN.
Vocera supports up to 128-bit Wired Equivalent Privacy encryption, and we couldn’t tell that WEP affected traffic, throughput or signal quality. The badges worked at the same range as an 802.11b-equipped notebook. Similarly, signal dead zones stopped our badges from working in the same way a notebook would stop working. Voice latency was approximately 1 second, and we could adapt to this slight hesitation quickly.
Conclusion
The Vocera 1.02 offering has a “one-dot-zero” feel to it. Offsetting the product’s youth is a serious “wow” factor. While we couldn’t test a large density of badges under stressful circumstances, the basic Vocera technology could survive the stresses of modern business – as long as the underlying wireless LAN transport is stable. The youthfulness and buggy-ness might be best described as Vocera’s Captain Kirk stage.
Henderson is principal researcher for ExtremeLabs of Indianapolis. He can be reached at thenderson@extremelabs.com.
