Aloha
Simulation Validation
Theoretical results are based
on Poisson distributed packet
generation rates and uniform
station traffic. Figure 31 shows
the simulation results plotted
against the theoretical function
described earlier. Any noticable
deviation is the result of simulation
time limitations. The variable
G represents initial transmission
and retransmission attempts
per slot while
is described by the function:
The simulation results are
indicated by *'s and were obtained
using uniform station traffic
similar to that used to validate
TDMA.
Figure 1: Slotted
ALOHA theoretical and simulation
results.
Because access to vacant slots
is controlled using Slotted
ALOHA which was previously validated,
no additional validation is
necessary.
Simulation Execution Results
The Slotted ALOHA simulation
results can be seen in Figures
2, 3, 4, 5 and 6.
Compared to GTDMA the utilization
and delay results come up short,
however the flexibility of Slotted
ALOHA greatly exceeds that of
GTDMA or TDMA. With respect
to the station queue results
the following should be noted.
The collision of a packet is
not determined by a station
until it is read one propagation
delay after its transmission
on the rebroadcast. During this
time the packet must be kept
in some kind of hold queue buffer
until the success of the transmission
is known. This hold queue buffer
must be of sufficient size to
accommodate the number of packets
which can be transmitted during
one propagation delay. After
the result of the transmission
is determined, the packet is
either purged from the hold
queue or requeued in the transmission
queue. Since the state of the
hold queue is constant and the
effects of collisions are seen
in the size of the transmission
queue, the packets in the hold
queues are not represented in
the graphs. With that said,
the most remarkable statistic
depicted in the queue buffer
graphs is the almost non-existent
average. This indicates that
while these queue buffers do
exhibit some peaks, mostly due
to the burst factor, most of
the time they are empty.
Figure 3: Slotted
ALOHA Queue Activity for burst=1
Figure 4:
Slotted ALOHA Queue Activity
for burst=5
Figure 5:
Slotted ALOHA Queue Activity
for burst=10
Finally in the last graph,
a very strong relationship between
packet size and delay can be
seen; the smaller the packet,
the worse the delay. The intuitive
explanation is that smaller
packets represent smaller slots
and a greater opportunity for
collision.
Figure 6: Slotted
ALOHA Packet Size vs. Delay
The Utilization vs. Delay graph
shown in Figure 7 shows results
similar to those obtained using
GTDMA with some notable differences.
For lower utilizations, the
delay characteristic of GTDMA
is better due to the lack of
contention. But at higher utilizations,
Reservation ALOHA has better
delay results. This is due to
the ownership of the slot for
as long as needed. Obviously
this protocol favors stream
type messages as opposed to
single packet messages.
Another important advantage
is the lack of dependence on
station population. In other
words, stations could be added
or removed without affecting
the mechanics of the protocol.
The queueing characteristics
seen in Figures 8, 9 and 10
are almost identical to those
of GTDMA. This would seem to
indicate that the contention
at slot acquisition is offset
by the contiguous assignment
of that slot to the station.
Figure 7:
Reservation ALOHA Utilization
vs. Delay
Figure 8:
Reservation ALOHA Queue Activity
for burst=1
Figure 9:
Reservation ALOHA Queue Activity
for burst=5
Figure 10: Reservation
ALOHA Queue Activity for burst=10
Predictably, the same relationship
exists between packet size and
delay with this protocol that
existed with the Slotted ALOHA
protocol. Once again it would
seem that smaller packets and
thus smaller slot sizes present
more of an opportunity for collision
which results in poorer delays.
The difference here is that
the performance stabilizes at
a smaller packet size which
provides an opportunity for
reducing the queueing buffer
requirements.
Figure 11:
Reservation ALOHA Packet Size
vs. Delay at 500 Mb/sec
Reservation Aloha
This channel allocation scheme
divides the channel bandwidth
into slot sizes equal to the
transmission time of a single
packet. This again assumes that
the packet sizes are of constant
length. The slots are organized
into frames of equal size whose
length spans the length of one
propagation delay. A station
makes an implicit reservation
by successfully transmitting
in an available slot. After
a successful transmission the
station is guaranteed that slot
in succeding frames until it
is no longer required. A slot
becomes unused either by going
empty in the previous frame
or by collision in the previous
frame. The remaining stations
can then compete for unused
slots using Slotted Aloha. If
the frame size did not span
the propagation delay, a slot
could go unused successive times
before stations sensed its availability.
Figure 12 shows several frames
depicting the execution of this
protocol. In frame 1 two stations
(stations 2 and 5) collide in
an attempt to acquire the same
slot. They succeed in frame
2. The same scenario occurs
in frames 5 and 6 except the
collisions occur between stations
2 and 4, and 5 and 1 respectively.
Figure 12: Reservation
Aloha Protocol
Advantages:
· Efficiently handles
bursty data traffic.
Disadvantages:
· Inefficient for single
packet messages.
· Requires queueing
buffers for retransmission of
packets.
· Requires tracking
status of slots in previous
frame.
· If propagation delay
is large, frame size could be
excessive.
Because of Slotted ALOHA's
limited utilization potential,
various methods have been developed
which improve upon its efficiency.
One of these methods is known
as Reservation Aloha. The main
modification has to do with
slot ownership after a successful
packet transmission. With Slotted
ALOHA any slot is available
for use by any station regardless
of its prior usage. With Reservation
ALOHA the slot is considered
owned temporarily by the station
which used it successfully.
When the station is through
with the slot it simply stops
sending. An idle slot is available
to all stations on a contention
basis.
Given the requirement that
the frame size must span the
propagation delay, the number
of slots per frame varies with
the channel speed. Table 6 shows
the frame sizes in slots for
a variety of channel speeds
and packet sizes and a propagation
delay of 0.113 seconds.
Table 6: Frame
sizes for various channel speeds.
| Channel
Speed (Mb/sec) |
400 |
500 |
600 |
700 |
800 |
900 |
1000 |
1100 |
1200 |
1300 |
| .5
Mb Slot |
91 |
113 |
136 |
159 |
181 |
204 |
226 |
249 |
272 |
294 |
| 1.0
Mb Slot |
46 |
57 |
68 |
80 |
91 |
102 |
113 |
125 |
136 |
147 |
| 1.5
Mb Slot |
31 |
38 |
46 |
53 |
61 |
68 |
76 |
83 |
91 |
98 |
As can be seen, the number
of slots per frame for a 1 Mb
slot greatly exceeds the number
of stations in the Mars regional
network. For this reason, a
station is allowed to attempt
to access more than one slot
per frame as needed. Additionally,
a station will attempt to acquire
a slot with a probability proportional
to the length of its transfer
queue. For example, a station
in which the current transfer
queue length is 80% of it's
maximum would attempt acquisition
of the slot with a higher probability
than a station whose queue length
is 20% of it's maximum.
Slotted
Aloha
Pure Aloha
Aloha
and Network Stability
Aloha
Simulation & Reservation
Aloha Protocol
Slotted
ALOHA Simulation Parameters
ALOHA
PROTOCOL IN C - LANGUAGE
