A Parallel Non-Neural Trigger Tracker for the SSC

The Superconducting Super Collider (SSC) is a major project promising to open the vistas of very high energy particle physics. When the SSC is in operation, data will be produced at a staggering rate. Current estimates place the raw data coming out of the proposed silicon detector system at $2.5 x 10^{16$ bits/second. Clearly, storing all events for later off-line processing is totally impracticable. A hierarchy of triggers, firing only on events meeting increasingly specific criteria, are planned to cull interesting events from the flood of information. Each event consists of a sequence of isolated ``hits,'' caused by particles hitting various parts of the detector. Collating these hits into the tracks of the approximately 500 particles/event, and then quickly deciding which events meet the criteria for later processing, is essential if the SSC is to produce usable information. This paper addresses the need for real-time triggering and track reconstruction. A benchmarked and buildable algorithm, operable at the required data rates, is described. The use of neural nets, suggested by other researchers, is specifically avoided as unnecessary and impractical. Instead, a parallel algorithm, and associated hardware architecture using only conventional technology, is presented. The algorithm has been tested on fully scaled up, extensively detailed, simulated SSC events, with extremely encouraging results. Preliminary hardware analysis indicate that the trigger/tracker may be built within proposed SSC budget guidelines.