A scheme for the optimal shaping of multidimensional constellations is
proposed. This scheme is motivated by a type of structured vector qua
ntizer for memoryless sources, and results in N-sphere shaping of N-di
mensional cubic lattice-based constellations. Because N-sphere shaping
is optimal in N dimensions, shaping gains higher than those of N-dime
nsional Voronoi constellations can be realized. While optimal shaping
for a large N can realize most of the 1.53 dB total shaping gain, it h
as the undesirable effect of increasing the size and the peak-to-avera
ge power ratio of the constituent 2D constellation. This limits its us
efulness for many real world channels which have nonlinearities. The p
roposed scheme alleviates this problem by achieving optimal constellat
ion shapes for a given limit on the constellation expansion ratio or t
he peak-to-average power ratio of the constituent 2D constellation. Re
sults of Calderbank and Ozarow on nonequiprobable signaling are used t
o reduce the complexity of this scheme and make it independent of the
data rate with essentially no effect on the shaping gain. Comparisons
with Forney's trellis shaping scheme are also provided.