Improving Loss Resilience with Multi-Radio Diversity in Wireless Networks

1. Improving Loss Resilience with Multi-Radio Diversity in Wireless Networks Allen Miu, Hari Balakrishnan, C. Emre Koksal Presented by Yu-En Tsai (Slides partially from Allen Miu’s Mobicom presentation)

2. The problem • New wireless applications demand high performance • But: wireless channels are loss-prone • Interference • Noise • Attenuation • Multi-path • Mobility, etc…  Inconsistent and poor performance

3. Current solutions are inefficient for recovering losses • Coding (e.g., FEC) • Hard on highly variable channels • Retransmission • Wasteful: outage durations can be long (tens to hundreds of milliseconds) • Bit-rate adaptation • Hard on variable channels • Slows down other clients

4. Internet Today’s wireless LAN (e.g., 802.11) • Uses only one communication path • Clients choose the “best” AP May use only one path AP1

5. MRDC Internet AP2 Multi-Radio Diversity (MRD) – Uplink Today’s wireless LAN (e.g., 802.11) May use only one path • Allow multiple APs to simultaneously receive transmissions from a single transmitter 10% AP1 20% Loss independence  simultaneous loss = 2%

6. Are losses independent among receivers? • Broadcast 802.11 experiment at fixed bit-rate: 6 simultaneous receivers and 1 transmitter • Compute loss rates for the 15 receiver-pair (R1, R2) combinations • Frame loss rate FLR(R1), FLR(R2) vs. simultaneous frame loss rate FLR(R1 ∩ R2)

7. Individual FLR > Simultaneous FLR y = x FLR R1 R2 R1*R2 FLR(R1 ∩ R2)

8. Challenges in developing MRD • How to correct simultaneous frame errors? • Frame combining • How to handle retransmissions in MRD? • Request-for-acknowledgment protocol • How to adapt bit rates in MRD? • MRD-aware rate adaptation

9. Patterns CRC Ok 1100 0000 1 R1 1100 0000 -- 1100 0010 X 1101 1010 R2 1 1100 1000 X 0001 1010 1100 1010 0 1. Locate bits with unmatched value 2. Select bit combination at unmatched bit locations, check CRC O Corrected frame Bit-by-bit frame combining TX: 1100 1010 Combine failure Problem: Exponential # of CRC checks in # of unmatched bits.

10. Block-based frame combining • Observation: bit errors occur in bursts • Errors may be restricted to few blocks • Divide frame into NB blocks (e.g., NB = 6) • Attempt recombination with all possible block patterns until CRC passes • # of checks upper bounded by 2NB • Failure rate increases with smaller NB

11. Failure decreases with NB and burst size 1.0 Frame size = 1500B Probability of failure 0.8 0.6 NB = 2 0.4 NB = 4 0.2 NB = 6 … NB = 16 0 10 20 30 40 50 0 Burst error length parameter

12. Flawed retransmission schemes • Conventional link-layer ACKs do not work • Final status known only to MRDC • Two levels of ACKs are redundant • Cannot disable link-layer ACKs

13. DATA RFA DATA DATA MRD-ACK ACK Request-for-acknowledgment (RFA) for efficient feedback IP IP MRD MRD link link link MRDC

14. MRD-aware rate adaptation • Standard rate adaptation does not work • Reacts only to link-layer losses from 1 receiver • Uses sub-optimal bit-rates • MRD-aware rate adaptation • Reacts to losses at the MRD-layer Implication: First use multiple paths, then adapt bit rates.

15. Experimental setup ~20 m R2 R1 L • 802.11a/b/g implementation in Linux (MADWiFi) • L transmits 100,000 1,500B UDP packets w/ 7 retries • 802.11a @ auto bit rate (6, 9, 12, 18, 24, 36, 48, 54) • L is in motion at walking speed, > 1 minute per trial • Variants: R1, R2, MRD (5 trials each)

16. 18.7 Mbps 2.3x Improvement 8.25 Mbps MRD improves throughput Throughput (Mbps) R1 R2 MRD Each color shows a different trial

17. MRD maintains high bit-rate Fraction of transmitted frames Frame recovery data (% of total losses at R1) via R2 42.3% frame combining 7.3% Total 49.6% (Raw FLR was 35%) 0 6 9 12 18 24 36 48 56 Selected bit rate (Mbps) LOWER VARIANCE MEANS BETTER FOR TCP

18. Delay Analysis Fraction of delivered packets User space implementation caused high delay 0 10-3 10-2 10-1 1 10-4 One way delay (10x s)

19. Discussions • This paper takes advantage of path diversity • Frame losses are often path-dependent (multi-path fading), location-dependent (noise) • Statistically independent between different receiving radios • 2 radios (APs) are available within the sender’s coverage. • How to place the radios? Not too close and not too far • Cost reduced? Then why not use 3 or more APs? • MRD complements both ARQ and rate adaptation • ARQ-based retransmissions work well for losses in short time scales • Rate adaptation works well for losses in large time scales

20. Discussions (cont.) • It’s not clear in the paper that whether physical antenna diversity would be helpful. • Will the computation cost be a problem when we implement MRDC in a low-end platform (e.g. router)? • All experiments are done on desktop/laptop. • Convention: first adapt bit rates, then use alternative paths. • MRD: first use multiple paths, then adapt bit rates.

21. Summary • Design of Multi-Radio Diversity WLAN • Block-based frame combining • Request-for-acknowledgment protocol • MRD-aware rate adaptation • Analysis of block-based frame combining • Experimental evaluation • MRD reduces losses by 50% and improves throughput by up to 2.3x

22. Related work • Physical layer spatial diversity techniques • Antenna diversity • 802.16 MIMO/802.11n • Retransmission with memory [Sindhu ’77] • Opportunistic forwarding [Biswas ‘05][Jain ’05] • Bit rate selection (AARF, RBAR, MiSer, OAR, Sample Rate)