ATM networking was conceived some 20 years ago. ATM installations will reac
h their peak in the next decade. It is time to begin to consider what techn
ology should follow ATM and how best to get there. In this cuticle, we addr
ess some of the most important issues regarding recent advances in ATM rese
arch, including IP/ATM integration, connection admission control, signaling
, and active and programmable networks. We present several research efforts
conducted here at Cambridge University, illustrating possible evolutionary
directions for future ATM networks. (C) 1999 Elsevier Science B.V. All rig
hts reserved.