Session 2 - WLAN

Robust Rate Adaptation in 802.11 Networks

• Assumption: Rate adaptation plays a critical role in performance
• Related work in RA not 802.11 compliant
• Goals
  • Througput
  • Robustness
  • easy implementation
• Experimental platform with programmable AP
• Typical Guidelines

1. Lower rate on severe packet loss
   • 10 successive successes, raise rate
     • 71.5% failure
   • 2 successive failures, lower rate
     • 63% failure
2. Change rate based on deterministic patterns
3. use probe packets
4. ?
5. ?

• RRAA key components
  • short term stats
  • Adaptive RTS
• Robust to random channel loss and mobility
• Adaptive RTS (A-RTS) uses additive increase/multiplicative decrease
• Nice selection of compared protocols
  • Consistently performed better

IQU

• Scenario
  • WLAN at conference
  • 1000s of users, 100s of APs
  • Flash crowd
  • Heavy loads
• Case studies to understand heavy load conditions
• Metrics
  • Per user thruput

(Did you consider defining a minimal acceptable service rate, and a throughput.user metric?)

• Overhead index
  • Noted that control traffic >> user data
  • AP overloading
  • Limit users

• Basic idea
  • Limit users
    • Make others wait
• Params
  • Wait-period
  • Estimated-load
  • work-period
  • hold-time
• IQU is a framework to allow admins to implement various policies
• Benefits
  • Consistency
  • Awareness of connectivity
  • Power saving opps
  • Changes expectation and behavior

Distributed Channel Management in Uncoordinated Wireless Environments

• Unpredictable uncoordinated hotspot deployment
• Take-home points
  • Static channel assignments cause unfairness
  • Channel hopping can improve fairness
• More hotspots than channels
  • Users can generally not hop between hotspots
  • Need fair division among access points, not users
• MAXchop - distributed channel switching algorithm
  • Need to account for channel switch overhead
  • can take advantage of POC
• Evaluations
• ns2 simulation
  • Commodity algorithms
    • least congested channel search
    • MAXchop
  • Found improved fairness over a wide range of real topologies
• Linux APs
  • Good results as well.
• Question
  • Doesn't GSM already do this? How is this different
    • More challenges due to lack of coordination
      • Question asker was not impressed with answer

Session - Cellular Networks

Mitigating Attacks on Open Functionality in SMS-Capable Cellular Networks

• Tragedy of the commons
• Previous work did DoS using text messaging
• Large scale attack pattern

1. Recognition
2. Reconaissance
3. Expoit
4. Recovery/Fix

• This work lies between steps 2 and 3
• Preserve fidelity of both voice and SMS during targeted SMS attack
• Various techniques based on allocation of SSDCH
• No silver bullet answer

A3: Application-Aware Acceleration for Wireless Data Networks

• Experiments over various apps, networks, protocols, parameters
• Predict future requests, based on app behavior
• Prioritized fetching
  • Divide data into diffrent catgories
• Infinite Buffering
  • Prevent flow control from throttling
• Application-Aware encoding
• Question
  • This approach violates layering, so new apps will perform poorly

Metastability of CDMA Cellular Systems

• Context: guaranteeing QoS in CDMA cellular

TCP-Aware Resource Allocation in CDMA Networks