Directional Antennas in Mobile Ad Hoc Networks:

1
Directional Antennas in Mobile Ad Hoc Networks

• MAC and Routing Issues
• Prepared by
• Rima Khalaf
• Behnam Rezaei

2
Based on the work of

• N. Vaidya, Y.b. Ko, and V. Shankarkumar, Texas
  AM University Medium Access Control Protocols
  for Directional Antennas in Ad Hoc Networks
• A. Nasipuri et. Al , University of Texas, San
  Antonio A Medium Access Protocol for Ad hoc
  Networks with Directional Antennas
• A. Nasipuri et. Al, University of Texas, San
  Antonio On Demand Routing Using Directional
  Antennas in Mobile Ad Hoc Networks
• R. Bagrodia, J. Martin, A. Ren, M. Takai,
  Computer Science Department, UCLA Directional
  Virtual Carrier Sensing for Directional Antennas
  in Mobile Ad Hoc Networks

3
Why Directional Antennas????
4
SPATIAL REUSE
5
RTS/CTS Handshake in 802.11
A
B
C
D
E
RTS
RTS
CTS
CTS
DATA
DATA
ACK
ACK
6
RTS/CTS in IEEE 802.11

• RTS and CTS contain proposed duration of
• data transmission
• All in-range nodes MUST wait for this duration
  before transmitting
• Advantage Elimination of Hidden terminals
• Disadvantage Wastage of network capacity (D
  cannot send anything to E)

7
MAC Protocols for Directional Antennas can be
Classified in 2 Categories

• MAC Protocols for networks where the nodes
  possess location information (by use of GPS for
  instance)
• MAC Protocols where nodes do not have location
  information but depend on the Angle of Arrival of
  the Received Signal

8
Vaidya (et al.) Model

• Each node knows its exact location and the
  location of its neighbors.
• Each node is equipped with 4 directional
  antennas.

9
Vaidya Scheme 1

• The node uses a directional antenna to send the
  RTS (D-RTS), whereas CTS Packets are sent
  Omni-directionally.
• Data Packets and Acks are sent directionally
• Any other node that hears the CTS ONLY blocks
  the antenna on which the CTS was received.

10
Vaidya Scheme 1 Cntd.
11

• Other nodes CAN transmit
• A B ? C xD ? E
• A ? B C ? D xE
• All DRTS may not get an OCTS reply
• (D E in the above scenarios cannot send OCTS
  if anyone sends them a DRTS because one of their
  antennas is blocked)
• Control packets may collide
• A B- -gt C ? D E

12
Problem with Scheme 1

• What if A wants to talk to B? A has no way to
  know that B is busy

A possible scenario of collisions with DRTS
packets
13
Vaidya Scheme 2

• The node uses two types of RTS namely,
  Directional RTS and Omni-directional RTS
  according to the following rule
• I) if non of the directional antennas of the node
  are blocked the node will send an
  omni-directional RTS.
• II) Otherwise the node will send a DRTS provided
  that the desired directional antenna is not
  blocked. If the desired antenna is blocked the
  node will defer until that antenna becomes
  unblocked.
• The CTS,Data and ACK are the same as before.

14
Further Optimizations the Wait to Send Packet
15
Performance
Throughput in Kb/s
16
But what if we have no location information?
17
Directional Virtual Carrier Sensing (R. Bagrodia
et. Al)

• Main Advantage No additional hardware is
  required to get location information.
• Each node estimates the position of its neighbors
  by the Angle of Arrival information, i.e. by
  noting the antenna which received the highest
  signal.
• Operation can be summarized in 3 main processes

18
1. AOA(Angle of Arrival) Caching

• Each node caches an estimated angle of arrival
  from neighboring nodes even if the signal is sent
  to it or not.
• When that node has data to send, it searches its
  cache for an angle of arrival information, if the
  AOA is found, the node will send a directional
  RTS, otherwise, the RTS is send
  omni-directionally.

19
1. AOACaching (Cntd .)

• The node updates its AOA information each time it
  receives a newer signal from the same neighbor.
• It also invalidates the cache in case if it fails
  to get the CTS after 4 directional RTS
  transmissions.
• The CTS is sent directionally.

20
2. Beam Locking and Unlocking

• When a node gets an RTS, it locks its beam
  pattern towards the source to transmit the CTS.
• The transmitting node only locks its pattern once
  it received the CTS.
• Beam Patterns are formed at both the transmitter
  and the receiver to maximize signal power.

21
Example
A
B
B

• Node A has data to transmit to node B and finds
  an AOA field for B in its cache.
• The AOA field is currently a little outdated
  since B has moved since the last update.

22
Example (contd)

• Node B senses the RTS from node A, and then
  adapts its antenna pattern to maximize the gain
  for the signal coming from node A.
• Node A locks its antenna pattern after the
  reception of the CTS from B.

23
3. DNAV Setting

• DNAV (Directional Network Allocation Vector)
  instead of NAV used in the classical 802.11
• DNAV also contains direction information of
  neighboring nodes. It is updated each time the
  physical layer provides new information about the
  location of a neighboring node.

24
3. DNAV Setting

• DNAV reserves the channel only in specified
  range of directions.
• The algorithm selectively excludes directions
  included in DNAV for transmission in which the
  node may cause interference with other
  transmissions in progress.
• Meanwhile it allows transmitting frames along
  other free directions.

25
Performance
PDR and throughput of the network in the mobility
scenario (without the physical CS in the IEEE
802.11).
PDRs with and without the physical CS
26
Performance Cntd.
Effects of the physical CS in mobility scenarios
Mix of omni-directional and directional antennas
27
Performance Cntd.
  Table 1 Set of parameters used in the
simulation.
PDR and throughput of the network in the no
mobility scenario (without the physical CS in the
IEEE 802.11).
Set of parameters used in the simulation
28
Problems with Bagrodia et al.

• Basically this scheme suffers from the same
  problem as Vaidyas scheme. This is a result of
  the directional RTS,i.e. a node can not know that
  a neighboring node is busy if that node sent a
  directional RTS to another neighbor.

29
Routing with Directional Antennas
30
On Demand Routing with Directional Antennas
(Nasipuri et al)

• Motivation Trying to reduce the routing overhead
  associated with the flooding of Route Request
  Packets associated with on demand protocols such
  as AODV and DSR.
• Directional antennas are exploited to limit the
  flooding to a specific region of the network thus
  reducing routing overhead,

31
Underlying MAC Protocol

• There is no use in modifying a routing algorithm
  for directional antennas if the underlying MAC
  protocol itself was not modified.
• The MAC protocol used is very similar to IEEE
  802.11 but the RTS and CTS are sent
  omni-directionally, but data is sent
  directionally.
• 2 Routing Protocols were proposed

32
Routing Protocol 1

• If a node S has a packet to send to node D, it
  transmits a query packet on the directional
  antenna which it had been using earlier to
  communicate with D.
• Assuming that D has not moved too far since the
  last communication instant, the flooding is
  restricted to the region containing the last
  known location to D.

33
Routing Protocol 1 (contd)

• If D has moved out of range, and S did NOT
  receive a Route Reply from D after a suitable
  timeout period, then S will generate a Route
  Request on all antennas (Omni- Directional), this
  time flooding the whole network.
• Result a decrease in Routing Overhead at the
  cost of increased latency.

34
Routing Protocol 1 (contd)
35
Drawbacks of Protocol 1

• The angular span of the antenna in S that was
  used to transmit data packets on the first hop of
  the of the last valid route to D may not
  necessarily include D.
• This happens if the first intermediate node on
  the path to D and the node D do not lie within
  the angular span of the same directional antenna
  of S.
• Solution Routing Protocol 2

36
Routing Protocol 2

• At the end of every successful route discovery,
  the source records the directions of the antennas
  to be used on every hop of the newly discovered
  route.
• The information is made available to the source
  if each node on the route adds to the header of
  the Route Reply packet, the antenna identifier
  which it used to receive the packet while
  forwarding it back to the source.

37
Routing Protocol 2 (Contd)

• This allows the source to get a rough estimate
  about the angular location of the destination by
  simply counting the number of times each antenna
  was used on the route that has been found.

38
Routing Protocol 2 (Contd)
39
Performance Tradeoffs

• If directional route search was successful from
  the first attempt, then both protocols have the
  advantage of remarkably reducing routing
  overhead.
• However, as node mobility increases, the chances
  of finding a route from a directional search
  drastically decrease.

40
Performance Tradeoffs Cntd.

• Another disadvantage is that a directional route
  request may find not always find the shortest
  route.
• This happens if the destination is not included
  in the search zone but an intermediate node which
  received the request a route to the destination
  and positively responds.

41
Finally

• Comparisons, Performance Tradeoffs, Proposed
  modifications, and conclusions

42
Performance Comparison
43
Proposed modifications

• What seems interesting to improve here is routing
  based on the random algorithms and probability of
  giant component when performing random routing.
• It seems that directional antennas can be used to
  model biological inspired connectivity and
  routing.
• Better use of Route Requests with antenna
  direction information

44
Conclusion

• We have presented 4 different MAC protocols, and
  one routing protocol that take advantage of the
  spatial reuse capabilities offered by directional
  antennas.
• All of the discussed protocols deliver optimum
  performance under no mobility conditions. Their
  performance degrade drasticaly as mobility
  increases

45
Conclusions Cntd.

• The use of directional antenna in ad hoc and
  large scalable networks is still unclear due to
  technical and physical difficulties of
  implementing directional antennas on each node.
• Those protocols are designed to increase network
  throughput at the cost of some increased design
  complexity
• Which brings us back to the popular saying

46
There is no such thing as a free lunch!
47
THANK YOU!
48
Questions???