We carried out a survey of the site coordinators in a mature telemedicine network of about 200 sites. The site directory contained information about 221 videoconference facilities. There were 191 site coordinators in all (i.e. some coordinators were responsible for more than one site). Of the 221 sites, we were able to contact 87 on first attempt and 155 by the fourth attempt. Thus there were 66 sites (30%) which were not contactable. We asked each site coordinator to describe any videoconference activity that had taken place over the previous five working days. Of the contacted 155 sites, 78 reported some videoconference activity during the period in question. The total reported videoconference activity was 12,800 min during the one-week monitoring period, that is, an estimated 924 h per month. The most common categories of work were education or training (511 h) and management or administration (225 h), which between them accounted for 80% of all reported videoconference activity. Fifty of the 155 sites (32%) reported that the equipment was not located in an area suitable for patient consultations. In addition, 20 sites (13%) volunteered that their videoconferencing facilities were not in working order at the time of the survey. We did not ask this question in the survey, so that this result represents a lower bound for the true number of inoperable systems.

We carried out a retrospective review of the videoconference activity records in a university-run hospital telemedicine studio. Usage records describing videoconferencing activity in the telemedicine studio were compared with the billing records provided by the telecommunications company. During a seven-month period there were 211 entries in the studio log: 108 calls made from the studio and 103 calls made from a far-end location. We found that 103 calls from a total of 195 calls reported by the telecommunications company were recorded in the usage log. The remaining 92 calls were not recorded, probably for one of several reasons, including: failed calls - a large number of unrecorded calls (57%) lasted for less than 2 min (median 1.6 min); origin of videoconference calls - calls may have been recorded incorrectly in the usage diary (i.e. as being initiated from the far end, when actually initiated from the studio); and human error. Our study showed that manual recording of videoconference activity may not accurately reflect the actual activity taking place. Those responsible for recording and analysing videoconference activity, particularly in large telemedicine networks, should do so with care.

Dsparrnicnt 0 1 Elccircinic and Electrical Engineerin;Univcr\ity 01 Strathclyde Third generation mobile r~d i o syrtenis will be required to operate in a wide vanety of environments from indoor picocells to outdoor macrtxells In additlon the range of services accommodated will increase relative to existing systems. The implications are that a single a r interlace 15 unlikely to satisty overall demands and that the emphasis must therefore be on flexibility. This piper exanlines the prospects of packet transmission to provide the necessar-y flexibility and draws attentioii t o some of the issues which the in~pleincntation of pickct access r;li\eThe pnniary advantage of packet t r a n s n~i w o n I\ IhaI bantlwidlh nia!' bc aliimtcd on a demand basis. I n voice communications 11 packet transniission is restncted to periods of voice activity this technique offers a theoretical improvement over fixed-allocation TDMA syslcms by a ratio in excess of 2.1, depending on the vilice activity detector. Other pitenha1 advantages of packet uansmission include the service independent nature of transmission (voice, video, data) and the ability to accommodate variable hit rate semices.The two main categories of I-andnni accc's? ;iic ALOHA and Caiiicr Sciisc Multiple Acces<. in hoth categories mihile\ contt,nd to XCC"\ ths channel 31 evcr) packct aii-ival S t i i c l caiiier beiisi: cy\rcnr\ aic iiiil practical i n the radio environnicnt h t u u~~ of the 'hidtlcn-u\ci-' piohlcrii. HoweLw i t is ps