Analyzing Multi-Channel Medium Access Control Schemes With ALOHA Reservation Yunghsiang S. Han† , Jing Deng‡ and Zygmunt J. Haas♯ † Graduate Institute of Communication Engineering National Taipei University, Taiwan, ROC ‡ Dept. of Computer Science University of New Orleans, USA ♯ School of Electrical & Computer Engineering Cornell Univ., USA Analyzing Multi-Channel Channel MAC Schemes Deng, Han, and Haas Outline • Introduction • The MAC-m scheme • Calculating the throughput of the MAC-m scheme • Numerical and simulation results • Conclusions GICE, National Taipei University 1 Analyzing Multi-Channel Channel MAC Schemes Deng, Han, and Haas Introduction • Medium Access Control (MAC) schemes are used to control the access of active nodes to the shared channel. • Multi-channel: one control sub-channel and m data sub-channels. • The control sub-channel is used for reservation of access to the data sub-channels over which the data packets are transmitted, and such reservation can be done through the use of the RTS/CTS (Ready-To-Send/Clear-To-Send) dialogue. • Single-channel MAC scheme: MAC-0 • Modified split-channel MAC scheme (m = 1): contention resolutions take place on the control sub-channel in parallel with the transmission of data packets on the data sub-channel (MAC-1). GICE, National Taipei University 2 Analyzing Multi-Channel Channel MAC Schemes Deng, Han, and Haas MAC-0, MAC-1S, and MAC-2 MAC−0 w1 Contention Resolution w2 Contention Resolution 1 1 Æ1 RTS CTS DATA 2 2 RTS CTS MAC−1S Æ2 DATA w2 Contention Resolution MAC−1 Æ2 2 2 RTS CTS DATA GICE, National Taipei University 2 2 RTS CTS Æ2 DATA 3 Analyzing Multi-Channel Channel MAC Schemes Deng, Han, and Haas Network Assumptions • The network is fully-connected, i.e., all nodes are in the transmission range of each other. • The packet processing delays and the radio propagation delays are negligible. • The traffic generated by active nodes (including retransmissions) is Poisson with rate G. • The contention resolution on the control sub-channel is solved by a pure ALOHA technique. • We normalize all variables with respect to the control packet transmission time. • The control package length is fiex at 48 bits in simulations. GICE, National Taipei University 4 Analyzing Multi-Channel Channel MAC Schemes Deng, Han, and Haas The MAC-m Scheme • There is a virtual, distributed queue. • Every node with data packets to send can compete on the control sub-channel by sending RTS packet to its intended receiver, which should reply with a CTS packet. • When the CTS packet is received at the sender of the RTS packet, the sender/receiver pair becomes a winner of the competition. • The following are the rules that a winner of the competition follows: 1. If there is an available (idle) data sub-channel after the requests in the distributed queue are assigned data sub-channels, the winner of the competition transmits on the available data sub-channel immediately after the CTS GICE, National Taipei University 5 Analyzing Multi-Channel Channel MAC Schemes Deng, Han, and Haas packet is received. 2. If all data sub-channels are busy and the distributed queue is not full, the winner of the competition adds itself to the distributed queue and waits for an available data sub-channel. 3. If all data sub-channels are busy and the distributed queue is full, the winner of the competition drops its right to transmit and goes back to the competition. • After transmission on the data sub-channel, the sender needs to go back to the control sub-channel to compete for the right of transmission for the next packet. • Whenever a data sub-channel becomes available due to the end of its current data transmission, it serves one of the customers in the distributed queue. If the queue is empty, the data sub-channel becomes idle. GICE, National Taipei University 6 Analyzing Multi-Channel Channel MAC Schemes Deng, Han, and Haas Contention Resolution of the MAC-m • A contention resolution period (W ) begins on the control sub-channel after a successful RTS/CTS dialogue. • The contention period lasts until there is a successful RTS/CTS dialogue. Contention Period, W CTS RTS CTS • The average duration of the contention period is 1 W = −1 . Ge−2G GICE, National Taipei University 7 Analyzing Multi-Channel Channel MAC Schemes Deng, Han, and Haas Arrival Statistics • We approximate the arrivals of successful RTS/CTS dialogues on the control sub-channel as a Poisson process, with arrival rate of λ. • The average arrival rate of successful RTS/CTS dialogues can be determined as Ge−2G 1 = . λ= −2G W +2 1 + Ge • In the steady state, the throughput of the whole system is at most the capacity of the control sub-channel. GICE, National Taipei University 8 Analyzing Multi-Channel Channel MAC Schemes Deng, Han, and Haas Throughput of MAC-m (m = 1) • The throughput by (M/D/1/1 + q model) is Sm ρ 1 = · . r + 1 p(d) + ρ 0 • r is the ratio of the data rate of the control sub-channel to the data rate of the data sub-channel. • ρ = λkr is the utilization factor, where k is the ratio of data packet size to the control packet size. (d) • p0 is the steady state probability of the empty virtual queue at the departure instant of customer from the system. • Let πn , n = 0, 1, . . . , q, q + 1, be the steady state probabilities that there are n customers in the system at any instant of time. • The average arrival rate of customers actually entering the GICE, National Taipei University 9 Analyzing Multi-Channel Channel MAC Schemes Deng, Han, and Haas system is λe = λ(1 − πq+1 ). • Therefore, the effective utilization factor is ρe = ρ(1 − πq+1 ) . • The system will be stable when the effective utilization factor is less than 1. • Hence, to determine the maximum achievable throughput, we need to maximize throughput under the constraint that ρe < 1. GICE, National Taipei University 10 Analyzing Multi-Channel Channel MAC Schemes Deng, Han, and Haas Throughput of MAC-m (m > 1) • The throughput (M/M/m/m + q model) is Sm 1 = r+m m+q X min{n, m} · πn n=1 ! . • The effective utilization factor is ρe = ρ(1 − πq+m ) . • The system will be stable when the effective utilization factor is less than 1. • Hence, to determine the maximum achievable throughput, we need to maximize throughput under the constraint that ρe < 1. GICE, National Taipei University 11 Analyzing Multi-Channel Channel MAC Schemes Deng, Han, and Haas Throughput of MAC-m with m = 1 = q in the fixed-total-bandwidth scenario 1 0.9 0.8 Throughput of MAC−m, Sm 0.7 0.6 0.5 0.4 0.3 Ld=4096, Tobagi L =4096, pdf d Ld=4096, M/D/1/2 Ld=1024, Tobagi L =1024, pdf d Ld=1024, M/D/1/2 0.2 0.1 0 0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 Ratio of control/data sub−channel data rates, r GICE, National Taipei University 0.8 0.9 1 12 Analyzing Multi-Channel Channel MAC Schemes Deng, Han, and Haas Throughput of MAC-m with m = 1, Ld = 1024 in the fixed-total-bandwidth scenario 0.8 MAC−0 q=10 q=3 q=2 q=1 q=0 0.7 Throughput, S 0.6 0.5 0.4 0.3 0.2 0.1 0 0.2 0.4 0.6 0.8 1 1.2 Ratio of control/data sub−channel data rates, r GICE, National Taipei University 1.4 1.6 13 Analyzing Multi-Channel Channel MAC Schemes Deng, Han, and Haas Optimal throughput of MAC-m with m = 1 in the fixed-total-bandwidth scenario 1 Optimum Throughput of MAC−m, S m 0.9 0.8 0.7 0.6 0.5 0.4 Ld=4096 Ld=2048 Ld=1024 0 1 2 3 4 5 6 7 Size of virtual distributed queue, q GICE, National Taipei University 8 9 10 14 Analyzing Multi-Channel Channel MAC Schemes Deng, Han, and Haas Optimum throughput of MAC-m (Ld = 1024) in the fixed-total-bandwidth scenario 0.75 Optimum Throughput, Sm 0.7 0.65 0.6 0.55 0.5 0.45 m=10 m=5 m=3 m=2 0 2 4 6 8 10 12 14 Size of virtual distributed queue, q GICE, National Taipei University 16 18 20 15 Analyzing Multi-Channel Channel MAC Schemes Deng, Han, and Haas Throughput of MAC-m (q = m) with different m and Ld in the fixed-total-bandwidth scenario 1 0.95 0.9 Optimum Throughput, S 0.85 0.8 0.75 0.7 0.65 MAC−M, L =4096 d MAC−M, Ld=2048 MAC−M, Ld=1024 MAC−0, Ld=4096 MAC−0, Ld=2048 MAC−0, L =1024 0.6 0.55 d 0.5 0 10 20 30 40 50 60 70 Number of data sub−channels, m GICE, National Taipei University 80 90 100 16 Analyzing Multi-Channel Channel MAC Schemes Deng, Han, and Haas Throughput of MAC-m with q = m in the fixed-channel-bandwidth scenario (r = 1) 1 0.9 0.8 Optimum Throughput, S m 0.7 0.6 0.5 0.4 0.3 0.2 Ld=4096 Ld=2048 Ld=1024 0.1 0 2 5 10 15 20 Number of data sub−channels, m GICE, National Taipei University 25 30 17 Analyzing Multi-Channel Channel MAC Schemes Deng, Han, and Haas Throughput of MAC-m schemes (m = 3, q = 3) for variable data packet length (Simulation and Analytical results) in the fixed-total-bandwidth scenario 1 0.9 0.8 Throughput of MAC−m, S m 0.7 0.6 0.5 0.4 0.3 L =4096, analysis d L =2048, analysis d L =1024, analysis d L =4096, simulation d L =2048, simulation d L =1024, simulation 0.2 0.1 d 0 0 0.5 1 Ratio of control/data sub−channel data rates, r GICE, National Taipei University 1.5 18 Analyzing Multi-Channel Channel MAC Schemes Deng, Han, and Haas Throughput of MAC-m with different m and Ld (Simulation and Analytical results) in the fixed-channel-bandwidth scenario 1 0.9 0.8 Throughput, S m 0.7 0.6 0.5 0.4 0.3 L =4096, fixed length, simulations d L =2048, fixed length, simulations d L =1024, fixed length, simulations d Ld=4096, variable length, analysis Ld=2048, variable length, analysis Ld=1024, variable length, analysis 0.2 0.1 0 2 4 6 8 10 12 14 16 Number of Data Sub−channels, m GICE, National Taipei University 18 20 19 Analyzing Multi-Channel Channel MAC Schemes Deng, Han, and Haas Conclusions • Based on the pure ALOHA contention resolution technique on the control sub-channel, we have developed a queueing model to study the performance of the MAC-m schemes with fixed-total-bandwidth scenario and fixed-channel-bandwidth scenario. • In fixed-total-bandwidth scenario, our analysis shows that multiple channel MAC schemes are always out-performed by the corresponding single shared channel schemes, given that the propagation delays are negligible. • In the fixed-channel-bandwidth scenario, each sub-channel always has the same data rate. Our study shows that there is an optimum number of sub-channels in these MAC schemes. GICE, National Taipei University 20