The knockout switch architecture has been found attractive for large-s
cale switch implementations because of its satisfactory cell loss perf
ormance, with constant output buffer speed-up independent of switch di
mension. The per port hardware complexity of a knockout concentrator,
however, does grow linearly with the switch dimension. In the paper, s
everal approaches are investigated to reduce the complexity of the kno
ckout while retaining the cell loss performance. A bufferless hierarch
ical concentrator architecture with reduced hardware complexity is der
ived. The concentrator complexity can be further reduced by introducin
g buffers in the concentrator, and the trade-off is analysed. Furtherm
ore, output grouping may be applied in the buffered hierarchical conce
ntrator to reduce the per port complexity. Two large-scale switch desi
gn examples are derived using the proposed design approaches, producin
g a complexity reduction ranging from 1.2% to 89.7%.