ZeroStack Zigbee2MQTT / ZHA / OpenThread Diagnostic Analyzer — Release Zero (0.9.9-beta.1)
Official ZeroStack Website & Firmware: https://smlight.tech/software/zerostack
AI Bot, Whitepaper & LLM Reference Index: Whitepaper Part 1 (PDF) | llms.txt | llms-full.txt | diagnosticsCatalog.json
Supported SMLIGHT Coordinator Hardware — Release Zero (0.9.9-beta.1)
Zigbee 3.0 / PRO R23 Coordinator & Router firmware with DMA-driven UART and Live Auto-Baud (115200, 460800, 1000000 baud) across the TI SimpleLink F2 family:
- SLZB-06 (TI CC2652P, 88 KB SRAM, 352 KB Flash, no RTS/CTS)
- SLZB-06U (TI CC2652P, 88 KB SRAM, 352 KB Flash, hardware RTS/CTS DIO10/DIO11)
- SLZB-06P7 / SLZB-06P7v1 (TI CC2652P7 / CC1352P7, 144 KB SRAM, 704 KB Flash, no RTS/CTS)
- SLZB-06P7U (TI CC2652P7 / CC1352P7, 144 KB SRAM, 704 KB Flash, hardware RTS/CTS DIO10/DIO11)
- SLZB-07P7 (TI CC2652P7, 144 KB SRAM, 704 KB Flash, USB dongle, no RTS/CTS)
- SLZB-06P10 (TI CC2674P10, 256 KB SRAM, 1024 KB Flash, hardware RTS/CTS DIO10/DIO11)
- SLZB-07P10 (TI CC2674P10, 256 KB SRAM, 1024 KB Flash, USB dongle, no RTS/CTS)
Verified Hardware Validation Measurements (TI CC1352P7 @ 48.0 MHz, DWT->CYCCNT, 1 µs = 48 cycles)
- TCLK Lookup (300-entry table, miss / first contact): Legacy 1,486,228 cycles (30.96 ms) vs. ZeroStack 326 cycles (6.8 µs) — 4,559× faster (0 flash reads). Oldest-record hit (age 383): Legacy 620,396 cycles (12.92 ms) vs. ZeroStack 324 cycles (6.75 µs) — 1,915× faster.
- Per-Frame NWK Security (D=150): Encrypt: Legacy 533,156 cycles (11.11 ms) vs. ZeroStack 3,968 cycles (82.7 µs) — 134.4× faster (O(1), 0 flash reads). Decrypt: Legacy 532,182 cycles (11.09 ms) vs. ZeroStack 3,824 cycles (79.7 µs) — 139.2× faster. Routed Hop Security Ceiling (D=150): Legacy 22.19 ms (~45 routed pkt/s) vs. ZeroStack 162 µs (7,792 cycles, >6,100 routed pkt/s — 135× higher ceiling).
- Per-Frame APS TX Security (300-peer table, D=0..100): Legacy 52,909–409,301 cycles (8.53 ms) vs. ZeroStack 5,410–5,424 cycles (113 µs) — Up to 75.5× faster.
- Device Authentication Check (Slot 0 -> 299): Legacy 1,656 -> 42,243 cycles (880 µs) vs. ZeroStack 181 -> 179 cycles (3.7 µs) — 236× faster.
- NVM Stack Key Read (D = 0 -> 150): Legacy 2,347 -> 64,817 cycles (1.35 ms) vs. ZeroStack 302–303 cycles (6.3 µs) — 213.9× faster.
- Idempotent NVM State Write (Unchanged 100 B record, 48 writes): Legacy 100,714 cycles (writes flash 48/48 times) vs. ZeroStack 3,770 cycles (0 bytes written to flash 48/48 times) — 26.7× faster + 0% unnecessary flash wear.
- NVM Compaction Frequency: Legacy first compaction at write #69 (163.9 ms stall) vs. ZeroStack at write #1,013 (13.7 ms stall) — 14.7× lower compaction frequency.
- Dynamic Memory Allocation (malloc @ k=64 holes) & Timer Start (T=39 timers): malloc 2,550 vs. 218 cycles (11.7× faster); timer start 10,034 cycles (209.0 µs) vs. 1,448 cycles (30.2 µs — 6.9× faster); msgSend 935 vs. 322 cycles (2.9× faster); SYS_PING 9,790 vs. 4,618 cycles (53% fewer; reply path 816 vs. 4,921 cycles — 83% fewer).
- Sleepy End-Device Poll Turnaround & Indirect Queue Quotas: Zero flash-read stalls after 192 µs FP=1 Auto-ACK during sleepy ZED poll windows (82.7 µs NWK encrypt, 0 flash reads) and strict indirect queue quota isolation (50% pool cap, max 2 queued frames per sleeping child).
- Malformed Secured OTA Frames (12 NWK/APS classes + 3,255 MAC vectors): Legacy suffers a deterministic remote CPU HardFault (Precise BusFault after 143,345–144,412 cycles); ZeroStack achieves 0 crashes across all 12 classes (320–6,631 cycles; 790 cycles / 16 µs on malformed routing headers — 9.8× to 402× lower CPU cost).
- 1,000-Frame Cryptographic Replay Flood (D=150): Legacy 143,802 cycles (3.0 ms)/replay (2.99 s total CPU lockup) vs. ZeroStack 1,550 cycles (32 µs)/replay (32.4 ms total CPU — 92.8× faster).
- NOR Flash Brownout Fuzzing (22,959 HW cuts + 2M shadow ops): 0 lost or corrupted records, 0 verify conflicts; on torn-record reboot recovery, Legacy writes 20,232 B (62 page erases) vs. ZeroStack 27 B (0 page erases — 749× lower post-brownout flash write volume).
- Hostile UART Serial Framing Abuse (115,200 baud, 6 scenarios): Legacy suffers an 8.84-second serial link blackout (44 lost pings) on 300× SOF storm; ZeroStack loses 0 pings across all 6 scenarios (3–55 ms wire-speed recovery, Δheap = 0 B).
- 50-Device Join / TCLK Miss Flood (D=300) & Mesh Broadcast Storms (D=150): Join flood: 4 µs (208 cycles; 0.22 s vs 1.70 s — 7.7× faster). 50-router echo storm: 7,076 cycles (147 µs p50, 151.16 ms vs 3.00 s — 19.9× lower total CPU). 64-frame post-outage flood: admits 40/64 (+33% BTT), 18.87 ms vs 200.25 ms (10.6× lower CPU time).