1. Kunal Amarnani, Ayush Bhardwaj, Angad Kumar Kapoor & Tejas Pargaonkar
MULTIRADIO WLANS
Kunal Amarnani, Ayush Bhardwaj, Angad Kumar Kapoor & Tejas Pargaonkar

2. Context, Problem Statement, Motivation
Each version of is assigned a spectrum of frequencies for use.
The spectrum is divided into 20 MHz bands.
Each band represents an orthogonal channel with an associated throughput.
One wireless radio connects to one channel.

3. Context, Problem Statement, Motivation
Study the issues with using multiple wireless radios from a single node.
Measure the performance when multiple radios are used on each node.
Attempt to solve any issues discovered when implementing multiple radios.

4. Context, Problem Statement, Motivation
Determine why the aggregate throughput experienced when using multiple radios is not equal to the ideal expected throughput.

5. Key Project Milestones
Set up a network – Accomplished
Measure the aggregate throughput when utilizing all radios - Accomplished
Determine why the aggregate throughput is not equal to the expected throughput - Accomplished
Come up with solutions to solve the aggregate throughput problem - Accomplished
Implement the best solution - Not-accomplished

6. Set up the Network
Two computers were set up, each with a network card with 3 radios
Access point and client setup
Frequencies of 5.26, 5.28, and 5.3 GHz
IP addresses of x, x, and x

7. Commands used Iwconfig – Setup up radio configuration
iwconfig ath0 essid "first" freq 5.26G rate 54M -Server
iwconfig ath0 essid "first" rate 54M -Client
Ifconfig – Setup IP configuration
ifconfig ath netmask Server
ifconfig ath netmask Client
Iperf – Create UDP data streams. Measure throughput
iperf -s -u -B p i 1 -Server
iperf -c u -p b 65M -t 100 -i 1 -Client

8. Measure the Aggregate Throughput
All 3 access points were listening while the 3 clients simultaneously transmitted
UDP packets with 100 second intervals
Ideal
Throughput (Mbps)
Observed
33.6
9.65 32
1.49
38.7
37.1
Aggregate
104.3
48.24

9. Figure 1. Orthogonal Channel Model
The Problem
Radios that use orthogonal channels should not theoretically suffer from interference.
Radios that are physically close may experience interference due to power leakage.1
5.26 GHz
5.28 GHz
5.30 GHz
Figure 1. Orthogonal Channel Model
1. Atul Adya, Paramvir Bahl, Jitendra Padhye, Alec Wolman, and Lidong Zhou, “A Multi-Radio Unification Protocol for
IEEE Wireless Networks”.

10. CSMA/CA Sender’s Side Power leakage occurs at the sender radios.
Physical carrier sense detects power leakage as interference.
The channels do not transmit data simultaneously.
5.26 GHz
5.28 GHz
5.30 GHz
Figure 2. Power Leakage

11. CSMA/CA Receiver’s Side Power leakage occurs at the receiver radios.
ACKed packets for one channel leak into received data packets of another.
Data collision occurs

12. Recognizing the Problem
All three channels were analyzed simultaneously with Wireshark.
The analysis showed that the channels were not transmitting simultaneously.
Radio 1
Radio 2
Radio 3

13. Proposed Solution (1) Structure loosely based on CSMA/CA protocol
Assumptions:
Disabled CSMA/CA at sender side
Disabled ACK message at receiver side
Perfect packet synchronization
Ignoring multiradio hidden terminal problem

14. Proposed Solution (2) Setup: Wireless ad-hoc network
Multiple nodes with variable number of radios

15. Proposed Solution (3) Sender Receiver RTS Step 1: Sending RTS
RTS sent along channel 1
In the RTS, number of radios possessed by sender is stored
Info of all the sender side radio and associated channels in 1 RTS
Attempt to reserve multiple channels for transmission
Sender
Receiver
RTS
Radio 1
Radio 1
Radio 2
Radio 2
Radio 3
Radio 3

16. Proposed Solution (4) Sender Receiver CTS Step 2: Sending CTS
CTS sent along same channel as before
Informs sender of all the channel that are free for transmission sent
Sender
Receiver
CTS
Radio 1
Radio 1
Radio 2
Radio 2
Radio 3
Radio 3

17. Proposed Solution (5) Sender Receiver Data Data Data
Step 3: Data Transmission
All the channels that are identified as free for transmission are used simultaneously (possible due to disabling CSMA/CA)
Sender
Receiver
Data
Radio 1
Radio 1
Radio 2
Data
Radio 2
Radio 3
Radio 3
Data

18. Proposed Solution (6) Sender Receiver SACK
Step 4: Data Acknowledgement
Selective ACK Sent from receiver to send letting sender know of packet loss or next expected packet (possible due to disabling ACK)
SACK Sent along one path only
Sender
Receiver
SACK
Radio 1
Radio 1
Radio 2
Radio 2
Radio 3
Radio 3

19. Proposed Solution (7) Advantages of solution:
Can support variable number of radios
Can support sender/receiver pairs having different number of radios
Allows for maximum utility of network as all possible channels are utilized

20. Proposed Solution (7) S2 S1 R1 R24 S2 S1 R1 1 1 2 2 3 3 4 2 1 4 R2
Here we can see both pair of sender/receiver being able to transmit data even though they have a different number of radios

21. Proposed Solution (8) S2 S1 R1 R24 S2 S1 R1 1 1 2 2 3 3 4 2 1 4 R2
Here S1 & R1 can only utilize channels 1 & 3 since 2 is being used by S2 & R2

22. Proposed Solution (9) Sender Receiver RTS ? ?
Drawbacks: Multiradio hidden terminal problem
As RTS with channel 1,2 and 3 information is being sent across channel 1, channel 2 and 3 are idle and unaware of this CTS/RTS transaction, therefore another node in the network can randomly use them since they are unaware of the CTS/RTS agreement
Sender
Receiver
RTS
Radio 1
Radio 1 ?
Radio 2
Radio 2 ?
Radio 3
Radio 3

23. Proposed Solution (10) Sender Receiver Data ? Data ? ? Data
Multiradio wireless also manipulates itself when packets are not sent in perfect synchronization
Sender
Receiver
Data ?
Radio 1
Radio 1
Radio 2
Data ?
Radio 2
Radio 3
Radio 3 ?
Data

24. Conclusion
Multiradio networks have great potential to increase network throughput substantially
Main problem is the interference caused by the power leakage from radios in close proximity
Main problem in implementing proposed solution was multiradio hidden terminal problem
Unable to attempt milestone 5 due to lack of time

25. Questions?