Timing recovery for an ATM network is inherently different from its ci
rcuit-switched counterpart, There are two generic approaches to ATM ti
ming recovery, namely, nonsynchronous methods which rely on cell jitte
r filtering and synchronous methods which use a common reference clock
. The latter approach has better jitter/wander performance, This is pa
rticularly significant for circuit emulation of existing hierarchical
signals (e.g. DSn signals), which have stringent jitter/wander require
ments. This paper compares several methods for timing recovery network
environment and discusses concept, implementation, and performance of
three synchronous techniques: Synchronous Frequency Encoding Techniqu
e (SFET), Time Stamp (TS), and Synchronous Residual Time Stamp (SRTS).
Among these methods, it is concluded that the SETS method, invented b
y the authors, is the most efficient, SETS has been accepted by ITU-T
as the timing recovery standard for AAL-1 (ATM Adaptation Layer - Circ
uit Emulation). Practical implementation considerations of the SRTS te
chnique and its robustness against cell loss are also examined in this
paper, In addition, it is shown that a strong analogy can be drawn be
tween SRTS and conventional pulse stuffing synchronization techniques
so that the jitter performance for SRTS is comparable to that of the c
ircuit-switched network.