Semi E171
SEMI E171 specifies Predictive Carrier Logistics (PCL), a data model and exchange standard for forecasting carrier movement and AMHS congestion ahead of time, built on top of the carrier-state tracking E87's Carrier Management System (CMS) already provides. E87/CMS answers "where is this carrier right now, and what's its slot-map status" — a real-time snapshot. PCL answers a question CMS was never built to answer: given current carrier positions, queue depths, and dispatch patterns, where is AMHS congestion likely to occur in the next few minutes, and which carrier moves should be rescheduled now to avoid it.
A few things are worth knowing about why carrier logistics needed a predictive layer in addition to CMS's real-time tracking:
CMS tells a host what already happened to a carrier; PCL is about what's about to happen to the whole AMHS network, which is a fundamentally different kind of data and a fundamentally different kind of standard. E87's CMS model is inherently reactive — carrier ID, location, and slot-map state are facts about the present moment, useful for knowing exactly where a lot is right now but silent on whether five other carriers are about to converge on the same intersection or stocker bay in the next ninety seconds. A fab's AMHS is a shared transport network, and transport networks develop congestion the same way highway traffic does: from the accumulated effect of many independent movements converging on a bottleneck, visible in the trend well before it's visible as a stalled carrier. PCL exists because nothing in CMS's data model was designed to project that trend forward.
PCL's standardization matters because without it, congestion forecasting becomes exactly the kind of fragmented, vendor-specific logic that SECS/GEM standards exist to eliminate. Any sufficiently sophisticated AMHS controller can already implement some form of proprietary traffic prediction internally — this isn't a new idea so much as an un-standardized one. The problem is that a fab increasingly runs AMHS equipment from multiple vendors, and a predictive capability that only exists inside one vendor's proprietary controller can't be queried, trusted, or acted on by a host-level scheduling system managing carrier logistics across the whole fab. PCL gives that forecasting capability a common data model — the same kind of move this glossary has tracked repeatedly, where an informal vendor capability becomes an interoperable standard once enough fabs need it to work the same way everywhere.
PCL sits downstream of E87's CMS and alongside E88's stocker-equipment model and E82's interbay/intrabay AMHS model as part of the carrier-logistics family this glossary has already covered — CMS supplies the real-time carrier facts, E88 and E82 model the storage and transport equipment moving those carriers, and PCL is the layer that turns all of that raw state into a forward-looking congestion forecast. None of those earlier standards were built to predict anything; they report what is. PCL is the piece that takes those reports as its input and does something none of them do on their own: project where the network is headed and recommend a correction before congestion becomes a measurable delay.
Read E171 as the standard that closes the gap CMS's real-time model was never built to close: knowing exactly where every carrier is right now doesn't tell a host where the AMHS network is about to get congested, and PCL is what actually turns that raw carrier-state data into a standardized forecast a host can act on before the jam forms.