Time and space switches; STTS and TSST

Role of time switches

A time switch (time slot interchanger, TSI) moves PCM samples from one time slot to another on the same highway.

 incoming slots -> [speech memory] -> outgoing slots
   write in order     ^ read address
                [control memory]

• Incoming samples are written into a speech memory in slot order (sequential write) and read out in the order set by a control memory (random read), or the reverse.
• It connects channel $i$ to channel $j$ on the same highway with a delay of up to one frame (125 µs).
• Built from RAM, so it is cheap, and it is non-blocking for one highway; the size is limited by memory speed ($c$ slots × 2 accesses within 125 µs).

Role of space switches

A space switch is a crosspoint matrix (electronic gates) that connects an input highway to an output highway, keeping the same time slot.

• Each crosspoint is opened or closed every time slot according to its control memory, so it is shared by many calls (time-shared space switching).
• It enlarges the switch to many highways but cannot change time slots; used alone it blocks when two calls need the same output in the same slot.

How combinations enhance capability

Real exchanges have hundreds of highways, so time and space switches are combined in multi-stage networks:

• TS / ST (two stage): simple, but blocking is high.
• TST (time–space–time): incoming TSI moves the call to any free internal slot, the space stage crosses highways, the outgoing TSI moves it to the wanted slot. The number of alternative paths equals the number of internal slots, so blocking is very low and memory is cheap. Standard in large exchanges.
• STS (space–time–space): space stages spread calls over $k$ time modules; suits smaller systems.
• TSST / TSSST: two or more space stages in the middle allow very large numbers of highways (large transit exchanges) without huge single matrices.
• STTS: two time stages in the middle give more slot-changing freedom and paths between groups of highways, useful in modular/distributed designs.

Benefits of these combinations:

1. Scalability: tens of thousands of lines and trunks switched in one network by adding modules.
2. Low blocking: many alternative paths (internal slots × middle arrays); with time expansion near non-blocking.
3. Lower cost and size: fewer crosspoints; most switching done in cheap RAM.
4. Reliability: alternative paths and duplicated planes let traffic avoid faults.
5. Flexibility: supports any-to-any connection of 64 kbps channels, broadcast/conference connections and $n \times 64$ kbps data paths, which digital networks and ISDN need.