This paper introduces a new application of fuzzy-logic control (FLC) t
heory to the power control in a direct-sequence, code-division multipl
e-access (DS/CDMA) cellular system over the mobile fading radio channe
ls. Power control is essential in DS/CDMA to compensate for the differ
ing received powers due to both the slowly varying long-term and fast
varying short-term fading processes and co-channel interference. The c
onventional feedback power control algorithms allow the base station t
o send a power command to either raise or lower each user transmitting
signal power level by a fixed power step and then keep the received p
owers almost equal. The fixed-step approach is actually an integral co
ntrol whose power increment is determined according to the bang-bang-l
ike control policy. However, this control scheme suffers from poor sys
tem stability, large overshoot, and long rise time. To tackle this dif
ficulty, a fuzzy proportional-plus-integral (PI) control, whose input
variables are the received power error and error change, is introduced
to determine each user's transmitting power in order to maintain simu
ltaneously all users' signal power received at the base station nearly
equal and to achieve better system stability and control performance.
The derivation of the fuzzy PI control has been carried out by analyz
ing both the closed-loop steady state behavior and transient response
of the system with a priori knowledge of the dynamics of the CDMA mobi
le fading channels. In fuzzy control, linguistic descriptions of actio
ns in controlling a process are represented as fuzzy rules. This fuzzy
-rule base is used by an inference mechanism in conjunction with some
knowledge of the states of process in order to determine control actio
ns. These control actions would lead to the fast rise time, minimum ov
ershoot, and small root-mean-squared (rms) tracking error. Furthermore
, the additional advantages of fuzzy PI control over conventional cont
rol theories are increased robustness despite interference and the abi
lity to handle the time-delay process without system degradation since
there is usually a latency between each user and base station. Simula
tion results show that the fuzzy PI power control provides much smalle
r rms tracking error and better traffic capacity performance compared
with the fixed-step control, especially in poor co-channel interferenc
e conditions.