HIMAC

1. HIMAC High Throughput MAC layer Multicasting Jaya Goyal MSc. 1st Year

2. CONTENTS • Why HIMAC • Performance Limitations in IEEE 802.11 MAC protocol * Channel State Indifference * Demand Ignorance • Proposed solution is HIMAC * UCF * UNF • Working of HIMAC • Possible cases • Performance Evaluations • Conclusions

3. WHY HIMAC • Currently we have IEEE 802.11 protocol to provide multicasting at MAC layer • It doesn’t have specific mechanism for it • Do so by broadcasting at BTR

4. Performance Limitations in 802.11 • CHANNEL STATE INDIFFERENCE • DEMAND IGNORANCE

5. CHANNEL STATE INDIFFERENCE • Broadcasting (BTR) can be < highest acceptable rate for multicast neighbors • acceptable rate changes frequently as time varying channel • Hence best approach is to find rate before each transmission

6. CHANNEL STATE INDIFFERENCE(2)

7. Experimental Result of Channel Aware Transmission

8. DEMAND IGNORANCE • In 802.11,we multicast i.e. broadcast here irrespective of demand • If node has already overheard data, still data is send to it.

9. Example to show the problem

10. SOLUTION PROPOSED • HIMAC protocol • Main focus is to improve throughput • It uses two mechanisms to handle it: * Unary Channel Feedback(UCF) * Unary Negative Feedback (UNF)

11. UCF • Before data transmission, find highest acceptable rate using UCF • S broadcasts RTS packet ( which contains MAC layer multicast addresses ) • Each potential R will send UCF (encodes highest acceptable rate) • Hence rate selection before each data transmission

12. Example showing effect of UCF signal on rate

13. UCF mechanism properties OVERHEADS • RTS • UCF Packets BENEFITS • Sender knows there is at least one active receiver • Sender can use transmission rate higher than BTR

14. UNF • Only needed in multihop network • Solution to demand ignorance along with UCF • In this basically, receiver respond with UNF if have overheard else UCF

15. UNF handles two scenarios 1) no demand for packet  sender needs to drop packet 2) heavy interference  packet retransmission is attempted

16. Example Showing drop of packet

17. Working Of HIMAC • S broadcasts RTS before packet transmission • Receivers on receiving RTS, if overheard the packet respond with UNF else UCF(+ highest acceptable rate) • If S receives only UNF, drop packet • Else if even a single UCF, send data at rate received with UCF

18. Possible Cases Case1: S receives both UCF and UNF: • Means some receivers have overheard some not • So will send data at lowest rate in UCF • Information of UNF is wasted Case2: UCF received, no UNF: • same as above (all receivers want data)

19. Possible Cases(2) Case3: UCF not received , UNF received: • Means all receivers have overheard the packet • Drop it Case4: both not received: • Means no overhearing (no UNF) • Receiver not in a position to receive packet(no UCF) • Retransmit later

20. Implementation of Unary Feedback • Sometimes it is difficult to extract the minm channel rate information from combined feedback • Hence here they use OFDM technology to receive feedback from multiple users without overlapping

21. PERFORMANCE EVALUATION • End-to-end throughput of multicast sessions using MAODV can be inc up to 74% • Reduces end-to-end latency by factor of 56 • Here we will see simulation results on packet size, network load and number of receivers due to space constraints.( on x axis)

22. Metrics for Performance Evaluation * End-to-end throughput : avg number of packets received by each receiver per second * MAC Latency: avg latency of receivers receiving the packets at MAC layer * End-to-end latency: avg latency of receivers receiving the packets at network layer

23. Impact of packet size

24. Impact of Network Load

25. Impact of Number of Recievers

26. Conclusions • In IEEE 802.11, no provision of multicasting specifically • Hence a new protocol HIMAC to handle this issue. • It uses two unary signals UCF and UNF for the same • It has increased throughput and reduced latency.