Back to index
Multiple Access in Wireless Digital Networks
Norman Abramson, IEEE/U. Hawaii
One-line summary:
Spread ALOHA is equivalnt to CDMA with a common spreading code for all
users, so ALOHA techniques can be carried over to spread-spectrum systems.
Overview/Main Points
- Advantage to using a single code for all transmitters in a CDMA
network, esp. for small cell sizes and multiple access satellite
apps
- Choice of a multiple-access protocol depends on traffic
characteristics and state of the technology at deployment time:
- DAMA (demand assigned multiple access): users request on a
separate control channel; request protocol introduces
delay and just moves the multiple-access problem to the
(lower-bandwidth) request channel
- DAMA w/random access: e.g. INMARSAT uses pure
ALOHA for request channel. OK since allocation tends to
be long-lived; no good if transmissions are bursty or
short-lived
- Multiple access protocols
- Slotted or pure ALOHA. Efficiency (r):
eff. channel capacity divided by capacity of
a continuous channel with same power & bandwidth. For
ALOHA, r=.18, asymptotically optimal for the special case
of small values of thruput and S/N ratio.
- Spread spectrum: max channel capacity in bits per
Nyquist sample: C = .5
log (1+P/N) based
on Shannon & Nyquist relations. "Spread spectrum" means
C << 1.
- CDMA. Multiply channel signals by orthogonal set of
spreading signals; multiply by cx conjugate at receiver.
Requires multiple receivers at CDMA BS to demodulate
received signal.
- Qualcomm CDMA (IS-95 std). Spreading code is
dynamically assigned via separate ALOHA channel when call
request is made. Up to 64 codes can be active at once.
- We can choose same spreading code for all CDMA users and the
channel will still have multiple-access capability (spread ALOHA):
- Each subchannel's bits will be offset by a constant amount
g
from the previous subchannel's bits within the frame
- With k subchannels, prob. that 2 bits will not
overlap is (1-1/g)^k; then total traffic
G = k/g.
- Get the noise-immunity of spread-spectrum with the nice
queueing properties of slotted ALOHA.
- Previous studies: no compelling evidence that there is a clear
advantage for multiple-code CDMA systems, despite their
complexity
Relevance
We can get the noise immunity of spread-spectrum together with the
queueing behavior of TDMA, by implementing wideband CDMA with a single
spreading code for all users but offsetting the channel bit sequences in
time rather than by using orthogonal codes.
Flaws
It's not clear to me why the "special case" for which plain old slotted
ALOHA is asymptotically optimal is an important special case.
Back to index