Session 5: Wireless Networking 2

Scribe: nshillin@nd.edu

===================

In Defense of Carrier Sense
M.Z. Brodsky, R. T. Morris (MIT)
______________________________________

Wireless medium is semi-shared (sometimes independent so use concurrent transmission, sometimes close together so use multiplexed transmission)

Solution: carrier sense (CS)

Problem: what about devices which are not independent but not close together, ie, in between?

Building blocks
* network layout
* radio propagation
* estimated throughput

Layout – place senders at fixed location
Radio propagation – path loss, shadowing, fading
Throughput – function of SINR, use Shannon capacity

Adaptive bit rate(ABR) – solves problem of whether to use multiplexing or concurrency.  Change bit rate to determine multiplexing vs concurrency

Implementation: experiments with indoor testbed

Summary: senders agree with receivers about concurrency vs multiplexing, hidden and exposed terminals, shadowing have little negative effect, adaptive bit rate helps, fresh look at modeling can help balance out the idiosyncrasies of in wireless systems

Question: Your model addresses classical wireless networks, how is the model affected by power adaptation in modern networks?
Answer: depends on reason for power adaptation,  ABR better solution, hard to give black and white answer

Question: How often do the points above the line occur? (referring to Figure 11 in paper)
Answer: out of 100 cases only 3 cases of exposed terminals

Question: What bit rate terminals did you use that resolved the problem of hidden and exposed terminals?
Answer: ran each bit rate combination (answer unclear)

Question: Combined with yesterday’s presentation, have you solved the problem of one hop communication?
Answer: Yes – still seems like worthy area of research

Interference Alignment and Cancellation
S. Gollakota,  S. D. Perli, D. Katabi
____________________________________

Throughput in MIMO lans limited by antennas at the APs

If there is a client and an AP the packets can be obtained by solving a linear combination of the transmission.

Problem: Can we deliver more concurrent packets?

Solution 1: Let the APs coordinate over Ethernet, emulate 4-antenna AP by sending signal over Ethernet (naïve solution), high overhead

Solution 2: IAC – align P3 with P2 at AP1
AP1  decodes P1 to its its bits
AP1 broadcasts P1 on Ethernet
AP2 subtracts/cancels P1 -> decodes P2, P3
Overcomes antennas/AP throughput limit

Contributions:
First MIMO LANs to overcome antennas per AP limit
Synthesis interferes alignment and cancellation
IAC doubles throughput

Alignment works independent of modulation phase

m-antenna case: for large m, IAC doubles throughput

Single active client: cannot have more than 2 concurrent packets
IAC picks best antenna on the AP, higher diversity than current MIMO, leverages PCF mode

Implementation: 20 node testbed, all nodes within range of each other
2.1 gains for IAC, 1.8 gains for IAC without diversity (UPLINK)
1.5 gains for IAC, 1.4 gains for IAC without diversity (DOWNLINK)
Better throughput at low SNR

Question: What is the rate achieved on testbed
Answer: 1 Mbps

Question: What about intercarrier interference
Answer: interference cancellation not affected by carrier interference because they use equalization

Question: What is the cost of IAC?
Answer: cost not high

Question: What are the limitations of MIMO effect?
Answer: if antennas are pretty close there will be interference, same limitations as current MIMO
Question: How would it scale with higher bit rates?
Answer: It would scale.

 
DIRC: Increasing Indoor Wireless Capacity Using Directional Antennas
X. Liu, A. Sheth, M. Kaminsky, K. Papagiannaki, S. Seshan, P. Steenkiste
________________________________________________________

Wireless technology causes dense deployment of wireless devices

Three orthogonal approaches to avoid interference
* frequency (SMACK, SWIFT)
*time (CSMA, TDMA)
*space (interference alignment, directional antennas)

Goal: spatial reuse using directional antennas

Limitations of directional antennas in indoor environment
*indoor space is scattered
*line of sight may be blocked

Spatial reuse – leverage of multiple paths and obstacles to improve spatial reuse

Finding antenna orientations
*max cap (naïve)
*find solution cost same as MAX SNR
*coordination among antennas
*carrier sensing fails so use centralized controller, TDMA scheduling MAC

Experiments: 2 indoor testbeds, 3 directional APs, 6 omnidirectional clients

Directional improves 70-80% over omnidirectional antennas

DIRC operations
*collect measurements
*compute schedules
*transmit

Conclusion: directional antennas can be used indoors, practical and lightweight

Question: What happens when one antenna is running and the other is not?
Answer: there are heuristics to reduce this problem

Question: How long to handle drop to zero during mobility?
Answer: 240 ms for 3 directional APs, linear increase in overhead with increased number of APs

Question: Have you compared your work with other companies?
Answer: compared to other companies we use MAX SNR

Question: Conflict maps have more conflicts, how much benefit does platform give?
Answer: no answer

Question: Do you have fairness in your formulations?
Answer: benefits come from spatial reuse so everyone gets better

Question: How would you benefit from MIMO?
Answer: no scanning but others remain the same

Question: How does layout affect the system? How does the system scale to other layouts/environs?
Answer: we update the measurements every 5 seconds so changes in environment are detected