Algorithm A1 TimesNet-driven adaptive BFT Consensus Loop.
Require:  History $\mathcal{D}$, View v, Timeout ${\Delta }_v$, Leader ${\mathsf{\ell }}_v$
Ensure:  Next View $v+1$, Leader ${\mathsf{\ell }}_{v+1}$, Timeout ${\Delta }_{v+1}$
1: Phase 1: Normal Consensus Execution
2:Node i executes HotStuff($v,{\mathsf{\ell }}_v,{\Delta }_v$)	▹ Standard BFT
3:if Block $B_v$ is Committed then
4:    Extract latency vector $L_{obs}$ from $B_v$.Metadata
5:    $\mathcal{D}\leftarrow \mathcal{D}\cup \left\{L_{obs}\right\}$	▹ History sync, Equation (1)
6: else
7:    Trigger ViewChange	▹ System Reset due to High Entropy
8: end if
9: Phase 2: TimesNet Inference
10:if  $\left|\mathcal{D}\right|\ge W_{\mathrm{warmup}}$  then
11:    for each node $j\in \mathcal{N}$ do
12:        ${\widehat{L}}_j\leftarrow \mathrm{TimesNet}(\mathcal{D}{|}_j,\theta )$	▹ Predict, Equation (2)
13:        $R_j\leftarrow {\widehat{L}}_j$	▹ Assign Robust Metric
14:    end for
15:    $\mathit{IsReady}\leftarrow \mathbf{True}$
16: else
17:    $\mathit{IsReady}\leftarrow \mathbf{False}$	▹ Insufficient history
18: end if
19: Phase 3: Dynamic Parameter Modulation
20:if not $\mathit{IsReady}$ then	▹ Cold Start Mode
21:    ${\mathsf{\ell }}_{v+1}\leftarrow ({\mathsf{\ell }}_{v}modN)+1$	▹ Fallback: Round-Robin
22:    ${\Delta }_{v+1}\leftarrow {\Delta }_{\mathrm{static}}$	▹ Fixed Timeout
23: else	▹ AI-Driven Mode
24:    $j^{*}\leftarrow arg{min}_{j\in \mathcal{N}}{R}_j$
25:    $\mathrm{Degrade}\leftarrow (R_{{\mathsf{\ell }}_v}-\mathrm{EMA}\left(R_{\mathsf{\ell }}\right))/\mathrm{EMA}\left(R_{\mathsf{\ell }}\right)$
26:    // 1. Leader Selection Logic
27:    if $(R_{{\mathsf{\ell }}_v}-R_{j^{*}})/R_{{\mathsf{\ell }}_v}>\delta$ or $\mathrm{Degrade}>\zeta$ then
28:        ${\mathsf{\ell }}_{v+1}\leftarrow j^{*}$	▹ Proactive Entropy Reduction, Equation (5)
29:    else
30:        ${\mathsf{\ell }}_{v+1}\leftarrow {\mathsf{\ell }}_v$	▹ Keep Incumbent
31:    end if
32:    // 2. Timeout Calibration Logic
33:    if $v(mod{W}_c)=0\vee {\mathsf{\ell }}_{v+1}\ne {\mathsf{\ell }}_v$ then
34:        ${\Delta }_{\mathrm{raw}}\leftarrow \alpha ·R_{{\mathsf{\ell }}_{v+1}}+(1-\alpha )·{\Delta }_v$	▹ Equation (6)
35:        ${\Delta }_{v+1}\leftarrow \mathrm{Clip}({\Delta }_{\mathrm{raw}},{\Delta }_{min},{\Delta }_{max})$	▹ Cooling-off
36:    else
37:        ${\Delta }_{v+1}\leftarrow {\Delta }_v$
38:    end if
39: end if
40:return  $v\leftarrow v+1,{\mathsf{\ell }}_{v+1},{\Delta }_{v+1}$