S-38.3310 Thesis Seminar on Networking Technology
The Improvements in Ad Hoc Routing and
Network Performances with Directional Antennas
Hao Zhou
Supervisor: Prof Sven-Gustav Häggman
Helsinki University of Technology
Communications Laboratory
8.8.2006
Agenda
 Introduction
• Research problem
• Objective and methodology
• Thesis roadmap
• Basic of smart antennas
 MAC protocol issue
• IEEE 802.11 MAC protocol
• Directional MAC problems
• Directional MAC proposals
 Routing protocol issue
• Ad hoc routing protocols
• Directional routing problems
• Directional routing proposals
 Case study
 Conclusion and future work
Introduction
 Ad Hoc Network can be deployed immediately on demand by surrounding
nodes without any fixed infrastructure supporting
 Each node in the ad hoc network is not only a host taking charge of
sending and receiving packets but also a router with responsibility for
relaying packets for other nodes
 Demand scenarios for ad hoc networks:
• Military environment
• Emergency situation
• Wireless sensor networks
• Low cost commercial communication networks
Research Problem
 A tremendous amount of MAC and routing protocols have been developed
for a hoc network where devices equipped with omni-directional antennas
 With fast development of smart antenna technology, directional antennas
have been proposed to improve ad hoc routing and network performance
 Several challenges and design issues arise when applying directional
antennas to ad hoc networks
Objective and Methodology
Objectives
 Introduce the smart antenna technology
 Discuss the MAC and routing problems of utilizing directional antennas
in ad hoc networks
 Survey directional MAC and routing proposals
 Evaluate routing and network performance between omni-directional
antennas and directional antennas in case studies
Methodology
 Literature study based on research papers, lecture slides, standardized
technical specifications
 Computer simulations with QualNet simulator
 Discussion with researchers working on ad hoc network studies
Thesis Roadmap
Chapter 3
MAC protocols
Chapter 4
Directional MAC proposals
Chapter 2
Chapter 7
Smart antennas
Case studies
Chapter 6
Directional routing proposals
Chapter 5
Routing protocols
Basics of Smart Antennas
 The smart antenna consists of multiple elements in a special configuration and
connected through complex weights. Smart antennas enable transmit and receive
with more energy in certain direction
 Switched beam antennas explore multiple fixed beams
in predetermined directions at the antenna site
 Adaptive array antennas could steer the main lobe towards
receiver in any direction dynamically
 Some advantages of directional antennas compared with omni-directional antennas:
•
•
•
•
They could reach large range with the same power due to higher gains
They could increase the channel capacity by rejecting interference better
They could alleviate multi-path effect by proving spatial diversity
They utilize power more efficiently.
Agenda
 Introduction
• Research problem
• Objective and methodology
• Thesis roadmap
• Basic of smart antennas
 MAC protocol issue
• IEEE 802.11 MAC protocol
• Directional MAC problems
• Directional MAC proposals
 Routing protocol issue
• Ad hoc routing protocols
• Directional routing problems
• Directional routing proposals
 Case study
 Conclusion and future work
IEEE 802.11 MAC Protocol
 IEEE 802.11 Distributed Coordinate Function (DCF) is developed to
provide communications between multiple independent mobile node pairs
without using access point or base station
 It utilizes Virtual Carrier Sensing (VCS) to alleviating collision happens in
channel
DIFS
Source
Node
RT
S
data
SIFS
Destination
Node
CT
S
SIFS
SIFS
t
AC
K
t
DIFS
NAV (RTS)
NAV
data
NAV (CTS)
t
Access to medium deferred
ACK: Acknowledgement
DIFS: Distributed Inter-Frame Space
RTS: Request to Send
backoff
CTS: Clear to Send
NAV: Network Allocation Vector
SIFS: Short Inter-Frame Space
Directional MAC Problems
 Neighbor location information and main lobe direction
The source node need to know the direction of destination node and neighbor
nodes in order to adjust the main lobe of antenna gain pattern for transmitting
 Extended transmission range
The directional data forwarding could reach beyond the reserved area with
conversional MAC protocol due to its higher gain
 New hidden terminal problem => Collisions
• Due to unheard RTS/CTS
The active node could not hear RTS/CTS send by other nodes due to directional
antennas has a larger gain in the desired destination than other directions
• Due to asymmetric gain
Node could not sense channel correctly with omni-directional antennas and
might interfere other on-going communications by directional forwarding packets
 Deafness problem
The source node fails to communicate with destination which is beam-forming to
another direction for on-going communication
Directional MAC Proposals-DMAC1/2
Directional MAC Scheme 1/2
Each node knows about location of neighbor nodes and itself based on GPS devices.
DRTS
DRTS
OCTS
E
ORTS
(DMAC2)
A
OCTS
B
DATA
ACK
OCTS
C
D
DATA
ACK
 DMAC 1 allows source node sends RTS directionally and receiver sends CTS
omni-directionally after receiving this RTS
(Node E is a potential interferer to on-going communication between Node A and B)
 DMAC 2 setting a condition before source node sending RTS:
• If none of the directional antennas of source node are blocked by other on-going
communications, source node send RTS omni-directionally
• Otherwise, it send a directional RTS to the other directions which are not blocked
Directional MAC Proposals-MMAC
Multi-hop RTS MAC scheme
Each node is equipped with an omni-directional antenna together with a directional
antenna
Forward RTS
B
A
C
D
G
F
Neighbor nodes can be divided into two groups:
• Directional-Omni (DO) neighbor
It could receive a directional transmission packet
even it is in idle mode with omni-directional antenna
eg A and B
• Directional-Directional (DD) neighbor
It is able to receive a directional transmission only
when its directional antenna beam-forms the source
node for reception
eg A and F
The basic idea is that DO neighbors help to establish an DD link by informing the location
of source and destination node with Forward RTS packet
Node A sends a Forward RTS to the DO neighbors one by one until to Node F, then F will
send directional CTS to A to help establish the directional communication link between
DD neighbors A and F
Directional MAC Proposals-DVCS
Directional Virtual Carrier Sensing Scheme
DVCS selectively disables particular directions including in which the node
would interfere with on-going communications and allows the node to
transmit to other directions, which increases the capacity greatly
New features :
 AOA caching
Every node estimates and caches the angle of
arrival of any signal received from its neighbors.
D
DNAV
A
C
B
DNAV
 Beam locking and unlocking
The node could lock its antenna pattern in the
directions of source and destination and unlock
after a successful packet transmission.
 DNAV setting
DNAV defines which angle range of the directional
antenna of that node should be disabled.
Directional MAC Proposals-C-DRTS
Circular DRTS scheme
Without using any predetermined neighbor location information, source
node uses all directional antennas circularly scanning the whole neighbor
area to inform the neighbor for intended communication
Traditional
Omni CTS
B
Communication
B
Communication
A
Traditional
DRTS
C
A
Circular
DRTS
C
Reception
Area
X
Each node has a location table which maintains the identity of detected neighbor, the beam
index on which it can be reached, the corresponding beam index used by the neighbor. It is
used for block beam directions that could produce inferences to active communication
Directional MAC Proposals-E-R/CTS
Extended RTS/CTS scheme
Each node knows about the neighbor node location information
Extended RTS
A
B
Added Lobe
C
Communication
F
E
☺
☻
Transmission
A
B
Communication
Three new features:
Reception
 Two lobe antenna pattern for DATA transmission
Source node sends a tone signal in the opposite direction of the active communication link
 Higher gain for RTS/CTS transmission
To overcome new hidden terminal problem due to asymmetric gain, the transmission range
of RTS/CTS is increased to cover the extra area caused by the DD link
 Transmission NAV and Receiving NAV setting
Different setting for transmission and receiving NAV to increase channel capacity, like Node
E and F could transmit in the directional of source node
Agenda
 Introduction
• Research problem
• Objective and methodology
• Thesis roadmap
• Basic of smart antennas
 MAC protocol issue
• IEEE 802.11 MAC protocol
• Directional MAC problems
• Directional MAC proposals
 Routing protocol issue
• Ad hoc routing protocols
• Directional routing problems
• Directional routing proposals
 Case study
 Conclusion and future work
Ad Hoc Routing Protocols
 Proactive routing protocol
• Maintain and update network topology knowledge for each node
• Utilize routing algorithm to exchange periodical link information
• High routing traffic and power consumption
• OLSR
 Reactive routing protocol
• Route discovery and route maintenance are on-demand
• Large delay but less routing traffic and less power consumption
• AODV
 Hybrid routing protocol
• Combine advantage of both proactive and reactive routing protocols
• High power consumption
• ZRP
Directional Routing Problems
 Directional route discovery problem
Current route discovery algorithms are carried out using an omnidirectional broadcast scheme, so DO and DD neighbor nodes which
could be reached by directional antennas are ignored
 Routing overhead problem
• One reason is that route discovery scheme broadcast route finding
packet omni-directionally
• Another reason is that some directional routing scheme produces
route redundant packets in route discovery procedure, like sweeping
scheme which sweeps the beam sequentially across all directions to
find the route
Directional Routing Proposals for directional route discovery
 Sweeping scheme
Through sequentially sweeping the antenna beam in omni-directional, DO
neighbors are easily detected, which leads to large routing traffic
 Heartbeat scheme
It could find the DO and DD neighbors by periodical broadcasting and
scoring of the heartbeat packets
• Informed discovery
After exchanging neighbor node information, each node
tries to directional transmit heartbeat packet to the twohop neighbors to establish DO link
• Blind discovery
With a synchronized time based on GPS devices, all nodes
performs discovery by a common direction which is
chosen by system. Each node alternates randomly between
sending heartbeat packets in that direction and listening in
the opposite direction to try to establish DD link
Blind discovery
Directional Routing Proposals for mitigating routing overhead
 Selective forwarding scheme
It prevent the same broadcast packet from transmitting
back to the node from which the packet is received
The intermediate node receiving the control packet will
forward it using half of its antenna beams in the opposite
direction of incoming angle of arrival
 Relay-node-based scheme
It innovates a manner to decide the relay node which could forward the control packet
efficiently and there is only one relay node in each of antenna element direction.
The node which is the farthest from the control packet sender is selected as relay node
 Location-based scheme
Each node obtain its location from a GPS device and attaches it in
the header of control packets. The receiving nodes will calculate the
additional coverage ratio and determine the forwarding delay, which
is inversely proportional to the additional coverage, for each
direction. The node must wait for the forwarding delay before
forwarding this packet. If same packet arrives within this forwarding
delay, the node will not forward in that direction.
Agenda
 Introduction
• Research problem
• Objective and methodology
• Thesis roadmap
• Basic of smart antennas
 MAC protocol issue
• IEEE 802.11 MAC protocol
• Directional MAC problems
• Directional MAC proposals
 Routing protocol issue
• Ad hoc routing protocols
• Directional routing problems
• Directional routing proposals
 Case study
 Conclusion and future work
Simulation environment parameters
 Routing and network performance comparison of directional antennas
with omni-directional antennas
 Simulation environment
• QualNet simulator
 Simulation results
• Throughput
• End to end Delay
• Packet delivery ratio
• Path length
The general simulation environment parameters
Parameter
Value
Propagation channel frequency
5 GHz
Path loss model
Two Ray
Directional antenna model
Switched beam
Directional antenna gain
15 dBi / 0 dBi
MAC protocol
IEEE 802.11 with DVCS
Directional NAV delta Angel
37 degree
AOA cache expiration time
2s
Element antenna pattern used in QualNet
Simulation Environment I-static communication distance case
Parameter
Value
Number of nodes
49
Node placement
Grid
Grid size
200 m
Terrain size
2000x2000 m
Simulation time
600 s
Bandwidth
24 Mbps
Transmission power
18 dBm
Receiver sensitivity
-83 dBm
Mobility model
None
Traffic type
Constant Bit Rate
Packet rate
8 packets/s
Packet size
512 byte
Number of flows
1
• The sender and receiver node place
between 7 different distance from 200 m
to 1400 m to see the network and routing
performance in static scenario in different
communication distance
Simulation Analysis I-static communication distance case
Throughput(kbits/s)
Throughput
34
AODV
AODV
OLSR
OLSR
32
30
(Omni)
(Dir.)
(Omni)
(Dir.)
28
200 400 600 800 1000 1200 1400
Distance between sender and
receiver(m)
• The end to end delay of AOVD with omnidirectional antennas increase more than OLSR
with the same antenna model; directiona antennas
have much better performance than omni-direcitonal
antennas; the increase of end to end delay much
depends on the increase of path leangth
End-to-End Delay
End-to-End Delay(ms)
• The throughtput of AODV and OLSR have no
big diffence in short communication distance;
the performance of AODV with omni-diectional
antenna decrease significently when the distance
is more than 1000 m; directional antennas are not
affected by increasing the communication distance
6
5
4
3
2
1
0
AODV
AODV
OLSR
OLSR
200 400 600 800 1000 1200 1400
Distance between sender and
receiver(m)
(Omni)
(Dir.)
(Omni)
(Dir.)
path length
Distance (m)
200
400
600
800
1000
1200
1400
AODV (Omni)
1
2
3
3
4
5
6
AODV (Dir)
1
1
2
2
2
3
3
OLSR (Omni)
1
2
3
3
4
5
6
OLSR (Dir)
1
2
2
2
2
3
3
Simulation Environment II-mobility speed case
• The Random Waypoint mobility model defines
three parameters: pause time; minimum speed
and maximum speed.
Parameter
Value
Number of nodes
50
Node placement
Random
Terrain size
1000x1000 m
Simulation time
900 s
Initial time
200 s
Bandwidth
24 Mbps
Transmission power
18 dBm
Receiver sensitivity
-83 dBm
Mobility model
Random Waypoint
Pause time
1s
Traffic type
Constant Bit Rate
Mobility level
1
2
3
4
5
Packet rate
4 packets/s
Minimum speed (m/s)
0
5
10
15
20
Packet size
512 bytes
Maximum speed (m/s) 5
10
15
20
25
• Each node randomly selects a destination location
within the physical terrain, and then it moves in
that direction in a speed uniformly chosen between
minimum and maximum speed. After it reaches the
destination, the node stays there for a pause time
period.
Simulation Analysis II-mobility speed case
Throughput
34
33
32
31
30
29
28
27
26
AODV(Omni)
AODV(Dir)
1
2
3
4
Speed level
5
Throughput(kbits/s)
Throughput(kbits/s)
Throughput
33
32
31
30
29
28
27
26
AODV(Omni)
AODV(Dir)
1
10 CBR
3
4
Speed level
5
30 CBR
• The throughputs of both antenna models decrease with
the increase of mobility level, but the throughput of
Throughput
Throughput(kbits/s)
2
33
32
31
30
29
28
27
26
directional antennas decreases slower than omnidirectional antennas
AODV(Omni)
AODV(Dir)
• With the increase of traffic load, the throughput of
1
2
3
4
Speed level
20 CBR
5
directional antennas doesnot have big changes, while
the one of omni-directional antennas decreases
accordingly
Simulation Analysis II-mobility speed case
End-to-End Delay
25
20
15
AODV(Omni)
AODV(Dir)
10
5
0
1
2
3
4
Speed level
5
End-to-End Delay(ms)
AODV(Omni)
AODV(Dir)
20 CBR
30
AODV(Omni)
AODV(Dir)
20
10
0
2
3
4
Speed level
5
• When the mobility level increases, the end to end delay
rises for both antenna models. In the heavy traffic load
scenario, the end to end delay increases slower than in
the other two light traffic load scenarios
35
30
25
20
15
10
5
0
3
4
Speed level
40
30 CBR
End-to-End Delay
2
50
1
10 CBR
1
End-to-End Delay(ms)
End-to-End Delay(ms)
End-to-End Delay
5
• The more traffic flows in the network, the larger is the
end to end delay
• The end to end delay of omni-directional antennas is
about four times of the one of directional antennas
Simulation Analysis II-mobility speed case
Packet Delivery Ratio
Packet Delivery Ratio
100
95
AODV(Omni)
AODV(Dir)
90
85
Packet Delivery
Ratio(%)
Packet Delivery
Ratio(%)
100
96
AODV(Omni)
AODV(Dir)
92
88
84
1
2
3
4
Speed level
5
1
10 CBR
2
3
4
Speed level
5
30 CBR
Packet Delivery Ratio
• The behavior of the packet delivery ratio is almost
the same as the one of the throughput
Packet Delivery
Ratio(%)
100
95
AODV(Omni)
AODV(Dir)
90
85
1
2
3
4
Speed level
20 CBR
5
• The directional antennas gain more than 7 % packet
delivery ratio when comparing with omni-directional
antennas
Simulation Analysis II-mobility speed case
Average path length
Average path length
2.4
2
1.8
AODV(Omni)
AODV(Dir)
1.6
1.4
1.2
number of hops
number of hops
2.2
2.2
2
AODV(Omni)
AODV(Dir)
1.8
1.6
1.4
1.2
1
2
3
4
Speed level
5
1
10 CBR
2
3
4
Speed level
5
30 CBR
Average path length
• The path length does have noticeable change when the
mobility increases or the traffic flow rises
number of hops
2.2
2
1.8
AODV(Omni)
AODV(Dir)
1.6
• This suggests that path length slightly depends on the
mobility speed level and traffic flows. The directional
antennas always save 25 % of the hops when
1.4
1.2
1
2
3
4
Speed level
20 CBR
5
comparing with omni-directional antennas.
Agenda
 Introduction
• Research problem
• Objective and methodology
• Thesis roadmap
• Basic of smart antennas
 MAC protocol issue
• IEEE 802.11 MAC protocol
• Directional MAC problems
• Directional MAC proposals
 Routing protocol issue
• Ad hoc routing protocols
• Directional routing problems
• Directional routing proposals
 Case study
 Conclusion and future work
Conclusion
 The network performance of directional antennas is not affected by increasing the
communication distance in static scenario
 The routing performance of OLSR outperforms AODV when devices equipped with
omni-directional antennas in long communication distance in static scenario
 The network performance deteriorates with increase of mobility level, but directional
antennas show significant advantage compared with omni-directional antennas.
 The important finding is that the network performance of directional antennas always
outperform omni-directional antennas both in static and mobility scenarios, and the
advantage of directional antennas is more obviously when channel condition
become worse or mobility level is large or traffic load is heavy
Future work
 This thesis concentrates on unicast routing protocol. The multicast routing protocol
is also an interesting issue that needs to be considered
 There is a need to implement a new directional route discovery algorithm for direction
antennas in the QualNet simulator to replace omni-directional route finding scheme in
order to mitigating broadcast storm problem
 The security is a very important issue in ad hoc networks. Since the ad hoc network
does not have any centralized control, the security must be processed in a distributed
manner