No Slide Title

Download Report

Transcript No Slide Title

Random Access
• ALOHA
• Carrier-Sense Techniques
• Reservation Protocols
• Voice and Data Techniques
- PRMA
- Adaptive CDMA
• Power Control
“Channel access algorithms with active link
protection for wireless communication
networks with power control”: Dan O’neill
7C29822.038-Cimini-9/97
ALOHA
• Data is packetized.
• Packets occupy a given time interval (TD technique)
• Pure ALOHA
– send packet whenever data is available
– a collision occurs for any partial overlap of
packets (nonorthogonal slots)
– Packets received in error are retransmitted
after random delay interval (avoids subsequent
collisions).
• Slotted ALOHA
– same as ALOHA but with packet slotting
– packets sent during predefined timeslots
– A collision occurs when packets overlap,
but there is no partial overlap of packets
– Packets received in error are retransmitted
after random delay interval.
Throughput*
• Throughput
– Measures fracture of time channel is used
– No power limitations
– Doesn’t measure true rate
• Assumptions
– Normalize slot time to 1
– Retransmission required for any packet overlap
– Infinite number of nodes
– Poisson packet arrivals at rate l.
• Slotted ALOHA
– For randomized retransmissions, the sum of
new and backlogged packet arrivals is Poisson
with rate G>l:
Throughput  GeG
• Pure ALOHA
Throughput  Ge2G
*Data Networks, 2nd Ed. Bertsekas and Gallager
S (Throughput per
Packet Time)
Throughput Plot
Ge G
.40
Slotted Aloha
.30
Ge2G
.20
Pure Aloha
.10
0
0.5
1.0
1.5
2.0
l
3.0
G(l) (Attempts per Packet TIme)
Note that there are two equilibrium points
for both slotted and unslotted ALOHA
• Comments
– inefficient for heavily loaded systems
– capture effect improves efficiency
– combining SS with ALOHA reduces collisions
Throughput with
Link Rates
Aloha Throughput (Abramson’94)
Cu Ge2G B log1  P /(GN ) 
r

C
B log1  P / N 
Efficiency
.8
-10 dB
P/N=-20 dB
.4
20 dB
0 dB
G
- Assumes power duty cycle is 1/G.
- High efficiency for low traffic and P/N
- Combines info. and queueing theory.
Spread Aloha

One CDMA code assigned to all users
t

Users separated by time of arrival
 Collisions occur when two or more
signals arrive simultaneously

Advantages
 Simplicity of transmitter/receiver
 No code assignment
 No limit on number of users for
sufficiently wideband signals (UWB)

Disadvantages
 Multipath can significantly increase prob.
of collisions
 RAKE harder to implement.
Carrier-Sense
Techniques
• Channel is sensed before transmission to determine
if it is occupied.
• More efficient than ALOHA  fewer retransmissions
• Carrier sensing is often combined with collision
detection in wired networks (e.g., Ethernet).
not possible in a radio environment
Busy Tone
Wired Network
Wireless Network
• Collision avoidance is used in current wireless LANs.
(WaveLAN, IEEE802.11, Spectral Etiquette)
8C32810.40-Cimini-7/98
Examples
• ARDIS
– slotted CSMA
• RAM Mobile Data
– slotted CSMA
• CDPD
– DSMA/CD - Digital Sense Multiple
Access
– collisions detected at receiver and
transmitted back
• WaveLAN
– CSMA/CA
8C32810.126-Cimini-7/98
Reservation Protocols
– A common reservation channel is used to
assign bandwidth on demand
– Reservation channel requires extra BW
– Offloads the access mechanism from the
data channel to the control channel.
- Control channel typically uses ALOHA
– Very efficient if overhead traffic is a small
percentage of the message traffic, and
active number of users small
– Very inefficient for short messaging
- For CDMA, reservation process must
assign unique spreading code to
transmitter and receiver.
7C29822.041-Cimini-9/97
Common Reservation
Protocols
• Demand–Based Assignment
– a common reservation channel is used to
assign bandwidth on demand
– reservation channel requires extra bandwidth
– very efficient if overhead traffic is a small
percentage of the message traffic
• Packet Reservation Multiple Access (PRMA)
– similar to reservation ALOHA
– uses a slotted channel structure
– all unreserved slots are open for contention
– a successful transmission in an unreserved
slot effectively reserves that slot for future
transmissions
7C29822.041-Cimini-9/97
Packet Reservation
Multiple Access
• Time axis organized into slots and frames
Frame 2
Frame 1
1
1
2
2
3,4
• All unreserved slots open for contention
•Transmit in unreserved slots with prob. p
• Data users contend in every slot (Aloha).
• For voice users, successful transmission in
an unreserved slot reserves slot for future
transmissions. Delayed packets dropped.
• Takes advantage of voice activity
(reservation lost at end of talk spurt).
PRMA Analysis

System states modeled as a Markov
chain.

Steady state probabilities used to
determine blocking probability.

Analysis complexity very high

Equilibrium point analysis (EPA) is
alternate technique



Equalizes arrival and departure rate for
any state
Used to obtain closed form solutions to
dropping probability.
Results match simulations well.
Performance

Reduces voice dropping probability
by 1-2 orders of magnitude over Aloha

User mobility




When a mobile changes cells, his
reservation is lost.
Delay constraint of voice may be exceeded
during recontention
Performance loss negligible
Bit errors



Voice bits received in error discarded.
Header bits received in error cause loss of
reservation
Nonnegligible performance impact
Dynamic TDMA

Frames divided into request, voice,
and data slots.

Voice slots reserved by voice users
using separate control channel.

Data slots dynamically assigned
based on pure ALOHA contention in
request slots.

Outperforms PRMA under medium
to high voice traffic.
Adaptive CDMA

CDMA uplink with synchronized users

Fixed chip rate Rc: spread signals occupy
bandwidth W

Voice and data users request service
from base station

Users admitted based on current
traffic, noise, interference, and type of
service request
Adaptive CDMA

SIR Requirements per user
Pu / Ru
  u,
P0 / W  ( Pt  Pu ) / W

u  v, d
Capacity constraint
Mv
Md
P0

 1
W
W
Pt  P0
1
1
Rv v
Rd  d






W: total spread bandwidth
Rv,Rd: symbol rate for voice,data
gv, gd: SIR requirement for voice,data
Mv,Md: number of users for voice,data
P0: Noise and out-of-cell interference power.
Pt=MvPv+MdPd: total power received at base,
where Pv is voice user power and Pd is data
user power.
Reservation Strategy

Voice nonadaptive: Pv, Rv, and v all
fixed.

Reserve some fixed number Kv
voice channels: maximum number is
dictated by capacity equation
K

max
v
P0
W

 1
Rv v
Pv
Adapt Md, Rd, and d to maximize
data throughput subject to capacity
constraint under active voice users.
Rate Control: Data

All data users admitted to the system

Variable rate transmission used to
maximize throughput given interference
from voice users

Variable rate transmission strategies:



Variable Bit Rate: users vary bit time Td=1/Rd.
Multicode: users assigned multiple
spreading codes, each modulated at fixed bit
rate Rd.
Variable Constellation Size: users assigned
one spreading code that is modulated with
variable-size constellations
Voice Call Blocking
Probability
• Voice blocking probability derived
from voice statistics and Kv
– Standard Markov analysis
lv/mv=10
lv/mv=1
lv/mv=5
Voice Occupancy: Kv/M0v
Average Throughput
Comparison
R0=50 Kbps, R0=100 Kbps
VBR/MC
VCS/MC
lv/mv=1
R0=100 Kbps
R0=500 Kbps
R0=50 Kbps
R0=500 Kbps
Voice call blocking probability
Analysis

Multicode has the worst throughput
 Codes interfere with each other

Variable bit rate outperforms
variable constellation size
 In VBR the bit rate increases
linearly with power
 In variable-rate MQAM the bit rate
increases logarithmicly with
power
 More efficient to vary the bit rate
than to vary the constellation size
 Variable bit rate may not be
practical
Throughput Gain with
Voice Activity
Detection
lv/mv=10
Pon=3/8
lv/mv=3
lv/mv=1
10-30
10-25
10-20
10-15
10-10
Voice call blocking probability
10-5
100
Power Control

Improves ALOHA efficiency

User with high power can capture a
packet even if there is a collision

Used in CDMA to maintain target SIR
of voice and data users

Can be used to maintain target SIR
for different user classes


Target SIRs must be feasible
Can combine with admission control to
maintain SIRs of active users
Main Themes

Retransmissions are power and spectrally
inefficient.

ALOHA has poor efficiency and does not
work well for data streaming

Reservation protocols are effective for long
data spurts but ineffective for short
messaging.

Voice and data supported by reserving
some channels for voice and using
remaining channels for (variable-rate) data

Power control can be used to maintain
QOS for all users in system – new users
blocked if degrade QOS for existing users
7C29822.042-Cimini-9/97