A　statistical　model　for　packet　transmission　probability　is　proposed　to　detect　network　bottleneck　link　in　the　field　of　statistical　inference　of　network　link-level　loss.　The　model　adopts　cumulant　generating　function　to　analyze　path-level　packet　transmission　probability　from　end-to-end　measurements　and　estimate　link-level　packet　transmission　probability.　Due　to　the　relationship　between　network　packet　transmission　probability　and　the　loss　probability,　the　link-level　packet　loss　information　can　be　indirectly　obtained.　Simulations　show　that　the　model　can　accurately　infer　network　link-level　packet　transmission　probability　information　without　high-load　probe　traffic　and　detect　network　bottleneck　link　in　terms　of　Chernoff　bound.