of the IEEE 802.11 Distributed Coordination Function Performance Analysis of the IEEE 802.11 Distributed Coordination Function (By Giuseppe Bianchi, 2000) Sunghwa Son
Outline Introduction 802.11 DCF Maximum & Saturation Throughput Performance Throughput Analysis Model Validation Maximum Saturation Throughput Performance Evaluation Conclusions
Introduction Distribution Coordination Function (DCF) Fundamental mechanism to access medium Based on CSMA/CA Wireless LAN uses half-duplex Describes two techniques Basic Access Mechanism RTS/CTS Mechanism
Basic Access Mechanism Channel Idle for a period of time Distribution Inter Frame Space (DIFS) Station generates random backoff interval Employs discrete-time backoff scale Transmission take at beginning of time slot Backoff time chosen between (0,w-1) Backoff time counter decremented to 0
Basic Access Mechanism
RTS/CTS Waits until channel sensed idle Transmits short frame called RTS Receiving station send CTS Carry information of length of packet Solves hidden terminal problems
RTS/CTS
Saturation Throughput Limit reached by system throughput as offered load increases Maximum load that the system can carry in stable conditions
Throughput Analysis Analytical Evaluation of Saturation Throughput Assumption Finite number of terminals and ideal channel conditions Fixed number of stations, each always having packet for transmission
Throughput Analysis Two cases Packet Transmission Probability Markov Chain model Throughput System throughput
Packet Transmission Probability Each packet collide with constant and independent probability p p will be referred to as conditional collision probability Model bidimensional process {s(t) , b(t)} with discrete-time Markov chain Stochastic Process b(t) representing backoff time counter Stochastic Process s(t) representing backoff stage
Markov Chain Model
Markov Chain Model Closed-form solution for Markov chain
Markov Chain Model Probability τ that a station transmits in a randomly chosen slot time
Markov Chain Model When m =0, i.e., no exponential backoff is considered, probability τ results to be independent of p In general τ depends on conditional collision probability p
Throughput Normalized system throughput S Probability that there is at least one transmission in the considered slot time Probability that exactly one station transmits on the channel, conditioned on the fact that at least one station transmits
Throughput Normalized system throughput E[P] is the average packet payload size Specify Ts and Tc to compute throughput for DCF access mechanism
Throughput Considering System via Basic Access mechanism Packet header H = PHYhrd +MAChrd Propagation delay δ
Throughput Packet transmission via RTS/CTS Access mechanism
Model Validation
Maximum Saturation Throughput τ depends on n (network size), W, and m. As n is not a directly controllable variable, the only way to achieve optimal performance is to employ adaptive techniques to tune the values m and W In the 802.11 standard, the values W and m are hardwired in the PHY layer details, and thus they cannot be made dependent on n
Maximum Saturation Throughput For n sufficiently large Results to be independent of n
Performance Evaluation Throughput of Basic Access mechanism depends on W Optimal value of W depends on n High value of W gives excellent throughput performance in case of 50 contending stations
Performance Evaluation Throughput obtained with RTS/CTS mechanism is almost independent of the value
Performance Evaluation Number of transmissions per packet increases as W reduces & network size n increases.
Conclusion Model suited for both Basic Access and RTS/CTS Access mechanisms Performance of Basic Access method depends on W and n Performance is only marginally dependent on the system parameters when the RTS/CTS mechanism is considered RTS/CTS scheme solved hidden terminal problem