I tested a 1.8 s group hang during a multi-agency pursuit at 19:40 and dropped idle grants by 21% while keeping late entry below 2%; the controller elevated command and EMS preempts automatically when queue depth rose. What values are you running at busy hour, and are you letting RSSI or grant queue depth drive timer shifts in real time?
I’ve had better results letting grant queue depth drive shifts with a bit of hysteresis: at busy hour I hold about 1.6 s and only drop to about 1.1 s after the queue stays ≤2 for 60 s, which keeps late entry around 1–2% and trims idle grants about 15–20%. RSSI is just a tie‑breaker near fringe sites, but without a 10 s minimum dwell it thrashes, so I’d avoid using it as the primary driver. Your “1.8 s group hang” during the 19:40 pursuit looks solid — if you keep that, consider pinning command at a 1.4 s floor whenever preempts are active.
Busy hour I hold about 1.5 s, but only drop to 1.0 s when grant queue ≤2 and inbound PTT stays under about 8/min; RSSI alone was too twitchy. For pursuits like your 19:40 test I add +200 ms on command/EMS preempt TGs to keep “late entry below 2%.” @vanessar88, have you tried PTT rate as the tie‑breaker instead of RSSI?
And nice result with “late entry below 2%” at 19:40. I get cleaner audio by driving shifts off the queue’s slope instead of the absolute level — if the queue keeps rising about 10 s I add about 150 ms, then only peel it back after about 30 s of flat, with a 1.2 s floor on interop talkgroups. Do you also freeze adjustments during affiliation bursts on the chase TGs?
For pursuits like your 19:40 run, I steer shifts off per‑site channel free‑time variance with a short EWMA of queue growth — RSSI’s too jumpy. Busy hour averages about 1.7 s here and only compresses toward about 1.2 s after about 30 s of stable variance; I also keep a higher minimum on patched interop groups so command/EMS boosts don’t thrash. @vanessar88 is right about trend, but is your controller’s “preempt” boost per‑site or system‑wide, and do you ease the timer when patch legs spin up?
I bias shifts off inter-PTT coefficient of variation and TG priority, not RSSI; when CV >0.6 I add about 250 ms and clamp a 45 s cooldown to stop flapping during pursuits like your 19:40 run with “1.8 s group hang”. Small caveat: when patches spin up, I cap a per-site minimum around 1.2 s or command gets starved despite preempts. @OP are you logging missed joins per TG or just the grant drop rate?
I’m at 1.6 s busy hour, bump to 2.2 s when the grant backlog climbs >2/sec and the median PTT gap drops <300 ms, with a 25 s hold so it doesn’t yo‑yo. I only let RSSI nudge the ceiling ±200 ms per sector when the noise floor spikes; otherwise it lies — still seeing ‘late entry <2%’. @jessiej do you weight backlog acceleration differently for priority TGs?
I’ve had better luck keying the bump off control‑channel BER instead of RSSI; when BER tops about 2% for 5 s and the last two floor changes land <350 ms apart, I add 200 ms (max +600 ms) and let it decay over 20 s. That’s kept “late entry” low without starving routine TGs, but I only allow the bump on TGs tagged pursuit/command to avoid bloating admin chatter. Does your controller expose BER or just RSSI and queue depth?