Communication load and delay in mobile trunked systems

This paper characterizes the traffic behavior of single channel conventional repeaters and various arrangements of trunked repeaters. The characterizations are based on extensive observations of repeater operation of each type. Key findings are: 1) Trunked system users talk longer than conventional repeater users (21.9 sec. vs. 13.7 sec. average message length); 2) Traffic loads from multiple sources tend to be uncorrelated and hence peak system load results from noise-like rather than direct additions of the individual source load peaks. An equation for estimating peak load under these conditions is presented along with comparisons of estimated and measured peak loads which confirm its utility; and 3) Erlang-C theory is applicable to the problem of quantifying trunked repeater performance. The message length cited in 1) favors the performance of conventional repeaters whereas the load-smoothing aspect noted in 2) favors the performance of trunked repeaters. A performance curve of average first access delay versus system air-time load in Erlangs is generated via Erlang-C theory for a Motorola trunked repeater system. This system utilizes a short time-out arrangement which substantially reduces the effective hold-time but appears message trunked to users due to priority queuing, hence first access delay is the relevant performance parameter. Fine-tuning the hold time to a value slightly greater than the average access duration produces a curve in good agreement with the performance observed on commercial systems.