RP operating frequencies are assigned
automatically and autonomously, eliminating the need for
manual frequency planning. The automatic frequency assignment
is called quasi- static autonomous frequency assignment
(QSAFA). QSAFA is a self-regulating means of selecting
individual RP frequency channel pairs that function without
a centralized frequency coordination between different
RPs. The QSAFA process is controlled by the RPCU for its
associated RP transceivers. To start the procedure, the
RPCU sends a message to a transceiver to turn off its
transmitter. The transceiver is instructed to tune to
the downlink frequency band and scan all possible downlink
frequencies. Then the transceiver reports the signal power
of the frequencies back to the RPCU and amplitude modulation
(AM). The frequency with the lowest received signal power
at the RP is selected. Finally, the RP transmitter turns
to the selected frequency and turns on.
The frequency assignment procedure is
repeated by all ports one at a time until no ports request
a change in their assigned frequencies for two consecutive
cycles. The procedure can be repeatetd until the algorithm
converges or until a threshold number of iterations is
reached. Because the downlink transmitter must be turned
off briefly during the measurement, the measurement should
be conducted during low-traffic hours. Simulation study
indicated that for 256 ports using 16 frequency pairs,
the assignments can always be stabilized within fewer
than five iterations. QSAFA combines the principle advantage
of dynamic channel allocation in that pre-engineering
of a frequency plan is unnecessary with the performance
advantages of a fixed frequency assignment; that is, elimination
of blind time slots for channel assignment, elimination
of call blocking due to resource blocking, and faster
call setup and handoff times.
Dynamic channel allocation for TDMA systems
is subject to blocking from two sources: interference
blocking, whereby the desired channel is blocked due to
interference; and resource blocking, whereby the desired
channel is blocked because the same time slot (not the
same channel) is already in use at the target RP. This
resource blocking probability seems to be higher than
Erlang-B blocking because of the blind time slot problem.
There are two aspects to this problem. The obvious one
is that SUs in handoff cannot see the time slot channels
they are using or adjacent time slots. The second problem
is that SUs attempting initial access or handoff may know
the target or best RP, but they do not know the traffic
pattern on that RP (i.e., which time slots may be in use
on other frequencies). As a result these SUs attempt access
to ports on time slots already in use, albeit on a different
frequency, and the RP cannot hear them. This problem can
be solved by the transmission of blind slot information
from each RP on the control channel, at the cost of reducing
alerting or system information capability.