Title: MACSCC: A Medium Access Control Protocol with Separate Control Channel for Multihop Wireless Network
1MAC-SCC A Medium Access Control Protocolwith
Separate Control Channel for Multi-hop Wireless
Networks
- Yijun Li, Hongyi Wu, Nian-Feng Tzeng,
- Dmitri Perkins, and Magdy Bayoumi
- The Center for Advanced Computer Studies (CACS)
- University of Louisiana at Lafayette
2Outline
- Introduction
- Motivation for this work
- Proposed MAC-SCC
- Experimental Results
- Conclusions
3Introduction
- Infrastructure-based network
- WLAN we are using
- Based on Access Points (AP)
- Ad-hoc network
- Infrastructure is not possible or expensive
- Military application, rescue, sensor network
4Ad Hoc Network-(multi-hop)
- Collection of self-configured nodes
- Each node works as host and router
Destination
Source
5Review of TCP/UDP
TCP
ACK 2
Timeout
Receiver
Sender
Sliding window Min( receiver window, congestion
window)
- Congestion Control
- Slow start for congestion window size
- Slow Start when timeouts. If timeout frequently
happens, TCP throughput will be low.
- Flow Control
- Receiver buffer not overloaded.
- Advertises available buffer size in ACK
UDP
- Whenever packet is ready, just send it
- Potential congestion without any control
- Unreliable without ACK.
6Review of IP layer
- TCP/UDP only consider sender and receiver. IP
layer will take care of how to make it
transparent in multi-hop network and instruct how
packets can send/receive between sender and
receiver. - Before a data packet from TCP/UDP can be sent
from source to destination, a route from source
node to destination should be discovered first.
Two type of routing schemes
- Proactive routing scheme
- Periodically send hello messages to direct
neighbors to maintain the topology information. - Two much control traffic.
- Reactive routing scheme
- Whenever source node wants to send a data packet,
it broadcasts route request packets to whole
network until destination is founded - Destination node sends back a route reply packet.
- The route information will be contained in the
packet header. - Need a long delay to find a route.
7MAC/PHY Layer
- MAC layer will try to avoid the collision of
access to the wireless media - PHY layer will transmit and receive bit-stream.
From hardware view, base-band processing and
frond-end transceiver
- 802.11x MAC
- 802.11 infrastructure-based and peer-to-peer
(ad-hoc) - 802.11eMAC Enhancements for QoS to improve QoS
for better support of audio and video (such as
MPEG-2) applications. - 802.11i Medium Access Method (MAC) Security
Enhancements enhance security and authentication
mechanisms. - 802.11x PHY
- 802.11 2Mbps (Proposed in 1997)
- 802.11b 1, 2, 5.5 and 11 Mbps, 100mts. range
(product released in 1999, no product for 1 or 2
Mbps) - 802.11g 54Mbps, 100mts. range (uses OFDM
product expected in 2003) - 802.11a 6 to 54 Mbps, 50mts. range (uses OFDM)
- 802.11n MIMO, LDPC and OFDM in PHY to increase
data-rate more than 100Mb
8Main problems for MAC design
- Hidden station outside senders transmission
range and in the receivers transmission range.
It will disturb senders transmission if it
transmits, due to it cannot detect the senders
transmission.
A
B
C
D
- Exposed Station in the senders transmission
range, but it misunderstand that it cant
transmit a packet
A
B
C
D
9RTS/CTS scheme for MAC in 802.11x
Send RTS, Reply with CTS
- Red machines receive RTS/CTS, get the period they
should keep silent. This period is NAV.
10RTS/CTS scheme for MAC in 802.11x
Send Data, Reply with ACK
After RTS/CTS reservation, red machines will keep
silent until the current data transmission is
finished. If red machines have packets to send
during reservation, they have to back-off
sometime after media is free to access
11Questions?
Is 802.11 MAC perfect?
NO
12Channel efficiency
NAV
Problem 1 in 802.11 MAC
13Link failure instability in TCP
- Collision in node C
- Several retries failed, report a link failure to
IP layer, then have to find a new route
Problem 2 in 802.11 MAC
14Unfairness in TCP
- In two-hop session, only communication between 4
and 5 can affect one-hop session. - In one-hop TCP session, the available interval
between packet transmission is larger than that
of the two-hop TCP session, which gives the
one-hop session more chances to transmit data. - Also, random back-off actually favors the last
succeeding transmission. - As the results, one-hop session will occupy the
entire wireless medium due to its unauthorized
priority
Problem 4 in 802.11 MAC
15Our proposed MAC-SCC
- The available bandwidth is partitioned into two
channels a data channel and a control channel,
each associated with a network allocation vector
(NAV). - The station transmits or receives on one channel
only at any given time. - During the current data transmission, the next
data frame can be pre-scheduled via the separate
control channel, and thus reducing the frame
collision probability and the bandwidth wasted
during back-off. - Moreover, the use of the separate control channel
helps to achieve fair medium access and solve the
instability problem resulted from frequent link
failures. - The optimal bandwidth partitioning between the
two channels is analyzed via a statistical model
16MAC-SCC protocol
- CH A Data channel, CH B Control channel
- S and D (enhanced RTS/CTS handshake)
- If CH A, CH B are idle, send RTS on CH A. If CH B
is idle only, send RTS on CH B. Otherwise,
back-off. - When RTS/CTS handshake occurs in CH B, send
SRTS/SCTS on CH A to reconfirm the reservation - D receives RTS/SRTS, replies with CTS/SCTS,
respectively - After handshake, S sends DATA, D replies with
ACK. - Other nodes
- contains NAVa, NAVb for CH A, CH B respectively.
- Update NAVa, NAVb
- When NAVa 0, convert NAVb into NAVa, release CH
B
17Example of MAC-SCC
No back-off here
DIFS
Pre-schedule the next data transmission
during the current data transmission
18Simplify hardware design
scheduling channel usage
- Two NAVs for two channels
- Listen to two channels, but only allow
transmitting or receiving at one channel at the
same time
19Further discussion on related work
20Issues in MAC-SCC
- How to partition two channels?
- What is system throughput?
- How about control packets overhead and link
failure? - How MAC-SCC affect TCP/UDP performance?
- Fairness in TCP
- UDP system throughput
- Two different ways to evaluate MAC-SCC
- Stand-alone simulation in Parsec
- Whole protocol stack simulation in Qualnet
21Analytical Bandwidth partitioning
Assume RTS arrival is Poisson distributed with
rate
Optimal D is 10
22Simulation Setup
- The traffic load (G) is defined to be the number
of frames per frame time - The bigger G, the more packets to send
23Stand-alone Simulations
- PARSEC A Parallel Simulation Environment for
Complex systems - Parsec is an parallel programming language based
on C. - We write PARSEC code to simulate and compare the
MAC-SCC protocol and 802.11 - System throughput, link failure probability, and
optimal bandwidth partitioning are studied
24Bandwidth Partitioning
- D Bandwidth of data channel / Bandwidth of
control channel - D gtgt 10, control channel ? bottleneck
- D ltlt 10, data channel ?bottleneck
- Optimal D is 10 and verifies the analytic results
25Throughput Comparison in Parsec
MAC-SCC keep flat in high traffic load
In high traffic load, MAC-SCC works much better
than 802.11 Gain is up to 60
26Link failure probability
Less failure in MAC-SCC
MAC-SCC always has lower link failure probability
due to scheduling packet
27Comprehensive Simulations
- Qualnet is a commercial simulator from
www.scalable-network.com - It contain whole protocol stack
- Deployed FTP sessions in application layer to
study TCP fairness - Deployed VBR in application layer to study
bandwidth partitioning under non-Poisson
distributed traffic - Used UDP to study system throughput in high
traffic load
28Fairness in two TCP sessions
Similar
- String topology, session 1 is one hop, session 2
is 1, 2, 3, or 4 hops - MAC-SCC get the similar throughput with the ideal
case - MAC-SCC total system throughput is a little bit
lower, due to TCP congestion control? no high
traffic load
29Fairness in Multiple TCP Sessions
Node Topology
- Study TCP fairness in more than 2 TCP sessions
(3, 4, and 8 TCP sessions) - Regular positioning to remove the effect from
node positions
30Three TCP sessions
802.11
MAC-SCC give fairness to node 5
MAC-SCC
X-axis is nodeID, Y-axis is system
throughput Node 2, 3 are one hop session Node 5
is two hop seesion
31Four TCP sessions
802.11
MAC-SCC
Node 2, 6 are one hop session Node 3, 8 are two
hop seesion
32Eight TCP Sessions
802.11
MAC-SCC
Node 2, 6, 11, 13 are one hop session Node
3,8,13,15 is two hop seesion
33Bandwidth Partitioning in VBR traffic
- We deployed VBR in application layers to study
bandwidth partitioning in non-Poisson
distribution with different traffic loads - D is between 8 and 12
34System throughput in high traffic load
- Scalar is used to control traffic load
- UDP is used for constructing high traffic load
35Conclusion
- We have proposed a novel Medium Access Control
protocol with a Separate Control Channel
(MAC-SCC). - To reduce hardware complexity, the station
transmits or receives on one channel only at any
given time. - The performance of MAC-SCC is quantified via
extensive simulations in both a stand-alone
simulator developed by using PARSEC and a
comprehensive network simulator called QualNet
with whole protocol stack. - Our results show that MAC-SCC can effectively
reduce the link failure probability, reduce
control packet overhead and transmission power,
achieve fair medium access when running multiple
TCP sessions, and yield a throughput gain up to
60 under high traffic load, when compared with
the basic RTS/CTS scheme.