Maximizing MAC Throughputs by Dynamic RTS-CTS Threshold MARCH 2004 Maximizing MAC Throughputs by Dynamic RTS-CTS Threshold Woo-Yong Choi and Sok-Kyu Lee Electronics and Telecommunications Research Institute Ladies and Gentlemen, My name is Woo-Yong Choi. I am working for Electronics and Telecommunications Research Institute. In this presentation, I want to talk about maximizing MAC throughput by dynamic RTS-CTS threshold. Woo-Yong Choi, ETRI

Contents Introduction Two Factors for Efficiency of RTS-CTS Method MARCH 2004 Contents Introduction Two Factors for Efficiency of RTS-CTS Method Dynamic RTS-CTS Threshold Control Method Throughput Analysis Optimal RTS-CTS Threshold Conclusions After a brief introduction,I will talk about two factors affecting the effectiveness of the transmission method using RTS and CTS frame exchange. And, then I will present the dynamic RTS-CTS threshold control scheme that can be implemented in wireless LAN systems. The numerical results of maximum MAC throughput and optimal RTS-CTS threshold for maximum MAC throughput will be presented. And finally, conclusions will follow Woo-Yong Choi, ETRI

Introduction TGn has the target of 100Mbps at MAC SAP MARCH 2004 Introduction TGn has the target of 100Mbps at MAC SAP MAC Protocol should be enhanced to meet this target RTS-CTS threshold can be dynamically adjusted to increase the MAC throughput More realistic assumption of data frame size distribution and hidden node situation is considered for the throughput analysis (cf. IEEE 802.11-03/0509r0) As you know, this task group has the target of 100Mbps at MAC SAP. For satisfying this target, we need to enhance MAC protocol. I think that one of methods for increasing MAC throughput is using dynamic RTS-CTS threshold according to the traffic state of BSS. At the last year, I presented about this topic. Today, I will present the new numerical results for maximum MAC throughput with more realistic frame size distribution and considering hidden node situation. Woo-Yong Choi, ETRI

Two Factors for Efficiency of RTS-CTS Method MARCH 2004 Two Factors for Efficiency of RTS-CTS Method Data Frame Length RTS-CTS method is efficient in the case of transmissions of long data frames But, inefficient with short data frames Numbers of STAs and Transmission Attempts Related with the number of retransmissions Efficient in the case of many STAs and transmission attempts But, inefficient with small numbers of STAs and transmission attempts There are two factors that can affect the efficiency of the transmission method using RTS and CTS frame exchange. One is data frame length, and the other is the numbers of STAs and transmission attempts. The latter factor of the number of STAs and transmission attempts is related with the number of retransmissions so that as the number of STAs and transmissions increase, the number of retransmissions will increases. The transmission method using RTS and CTS frame exchange is efficient for transmitting long data frames, but inefficient with short data frames. This is because the relative overhead of RTS transmission becomes larger with short data frame transmissions. The RTS-CTS threshold or dot11RTSThreshold in the specification is designed to consider this factor. In addition to this factor of data frame length, we propose that the numbers of STAs and transmission attempts affect the efficiency of the transmission method using RTS and CTS frame exchange. RTS-CTS method is efficient when the numbers of STAs and transmission attempts are large, that is, the number retransmissions is large. But, the RTS-CTS method is inefficient when small number of STAs and transmission attempts are involved with a BSS. This is because the transmission method using RTS and CTS frame exchange can reduce the number of the retransmissions of data frames by reserving channel through RTS and CTS frame exchange. For this reason, the RTS-CTS threshold should be optimally selected to maximize MAC throughput considering these factors. Woo-Yong Choi, ETRI

Dynamic RTS-CTS Threshold Control Scheme MARCH 2004 Dynamic RTS-CTS Threshold Control Scheme Infrastructure BSS AP periodically monitors the number of STAs connected to itself, the number of transmission attempts and the data frame size distribution AP calculates the optimal dot11RTSThreshold AP broadcasts the updated threshold value to be used by STAs Detailed algorithm is needed Independent BSS For further research We can design roughly the dynamic RTS-CTS threshold control method in infrastructure BSS. First, AP periodically monitors the state of its BSS, the number of STAs, the number of transmission attempts and data frame size distribution concerning the factors affecting the efficiency of RTS-CTS method. And, based on the updated state of BSS, AP optimally selects the value of RTS-CTS threshold. AP broad cast this updated optimal RTS-CTS threshold to STAs in its BSS. STAs will use the broadcast threshold value in determining whether the RTS-CTS frame exchange is used or not for their data frame transmissions. The detailed algorithm for calculating the optimal RTS-CTS threshold based on the state of its BSS is still for further research. And, the method for dynamically controlling RTS-CTS threshold is also for further research. Woo-Yong Choi, ETRI

MARCH 2004 Simulation Condition n greedy STAs attempt to transmit data frames continuously using DCF protocol The length of data frames is variable based on the experimental statistics from NLANR (National Lab. for Applied Network Research) (www.nlanr.net/NA/Learn/paketsizes.html) p: the probability that a transmission attempt fails due to the hidden node problem (p = 0, 0.25, 0.5) Optimal RTS-CTS threshold was obtained using computer simulations for maximizing MAC throughput We analyzed the optimal RTS-CTS threshold and the maximum MAC throughput using computer simulations.In computer simulations, n greedy STAs attempt to transmit data frames continuously using DCF protocol. The length of data frames is variable based on the experimental statistics from NLANR. We considered the hidden node proble using the probability p that a transmission attempt fails due to the hidden node problem. Optimal RTS-CTS threshold was obtained using computer simulations for maximizing MAC throughput. In a real situation, the algorithm for optimal threshold should be developed to implement the dynamic control scheme Woo-Yong Choi, ETRI

Payload Size Distribution MARCH 2004 Payload Size Distribution The frame size distribution is like this figure. Woo-Yong Choi, ETRI

Throughput Analysis (p=0) MARCH 2004 Throughput Analysis (p=0) This figure shows the numerical results of MAC throughputs using computer simulations when there is no hidden node, that is, p equals to 0. This line of rectangular points represents the maximum MAC throughput using optimal RTS-CTS threshold of IEEE 802.11a wireless LANs. And, this upper line represents the maximum MAC throughput of IEEE 802.11b wireless LANs. The other lines represent MAC throughputs of IEEE 802.11a and wireless LANs when RTS-CTS frame exchanges are always used or never used for every data frame transmission. Interestingly the minimum MAC throughput corresponded to one of the cases of the RTS and CTS frame exchanges always used or never used for data frame transmissions. Woo-Yong Choi, ETRI

Throughput Analysis (p=0.25) MARCH 2004 Throughput Analysis (p=0.25) This figure is for MAC throughput numerical results when p equals to 0.25. Woo-Yong Choi, ETRI

Throughput Analysis (p=0.5) MARCH 2004 Throughput Analysis (p=0.5) This figure is for MAC throughput numerical results when p equals to 0.5. Due to the effect of hidden nodes on MAC throughput, MAC throughputs become smaller as p increases. Woo-Yong Choi, ETRI

Optimal RTS-CTS Threshold (p=0) MARCH 2004 Optimal RTS-CTS Threshold (p=0) This figure is for optimal RTS-CTS threshold of IEEE 802.11a and b wireless LANs that maximize MAC throughput when there is no hidden node. When the number of STAs, n increases, the optimal RTS-CTS thresholds become smaller. This indicates that as the number of STAs increases, the RTS-CTS frame exchanges should be used more frequently to maximize MAC throughput. Woo-Yong Choi, ETRI

Optimal RTS-CTS Threshold (p=0.25) MARCH 2004 Optimal RTS-CTS Threshold (p=0.25) This figure is for optimal RTS-CTS threshold of IEEE 802.11a and b wireless LANs when p equals to 0.25. Woo-Yong Choi, ETRI

Optimal RTS-CTS Threshold (p=0.5) MARCH 2004 Optimal RTS-CTS Threshold (p=0.5) This figure is for optimal RTS-CTS threshold of IEEE 802.11a and b wireless LANs when p equals to 0.5. Woo-Yong Choi, ETRI

Performance Improvements MARCH 2004 Performance Improvements Average 25 % throughput improvement in IEEE 802.11a Average 28 % throughput improvement in IEEE 802.11b We observed that compared with minimum MAC throughput result by the worst case selection of RTS-CTS threshold, optimal RTS-CTS threshold increases MAC throughput by on the average 25% in IEEE 802.11a wireless LANs. And, optimal RTS-CTS threshold increases MAC throughput by on the average 28% in IEEE 802.11b wireless LANs. Woo-Yong Choi, ETRI

Conclusions Dynamic RTS-CTS threshold can improve MAC throughput MARCH 2004 Conclusions Dynamic RTS-CTS threshold can improve MAC throughput We analyzed the throughput with the consideration of realistic data frame size distribution and hidden node situation We propose that the value of dot11RTSThreshold used by (Q)STAs in the (Q)BSS be broadcast via a new information field in beacon frame We can conclude as follows. Dynamic RTS-CTS threshold can improve MAC throughput. We analyzed …. Woo-Yong Choi, ETRI