OAR: An Opportunistic AutoRate Media Access Protocol for Ad Hoc Networks B. Sadeghi, V. Kanodia, A. Sabharwal, E. Knightly Presented by Sarwar A. Sha 802.11b – Transmission rates Different modulation methods for transmitting data. – Binary/Quadrature Phase Shift Keying – Quadrature Amplitude Modulation Each packs different quantities of data into the modulation. The highest speed has most dense data and is most vulnerable to noise. Highest energy per bit 1 Mbps 2 Mbps 5.5 Mbps 11 Mbps Time Lowest energy per bit Transmission Throughput Why would a node ever want to slow down? – Longer transmission distance – More robust modulation – Moving node rapidly changes channel conditions Must adapt to channel conditions based on SNR Image courtesy of G. Holland Background IEEE 802.11 multi-rate Support of higher transmission rates in better channel conditions Auto Rate Fallback(ARF) – Use history of previous transmissions to adaptively select future rates – Error free transmissions indicates high channel quality – Lucent ARF implemention reduces rate after 2 lost ACKs, then attempts to speed up after a time interval Receiver Based Auto Rate (RBAR) – Use RTS/CTS to communicate a transmission rate based on channel quality. Receiver determines rate. Motivation Consider the situation below – ARF? – RBAR? B A C Motivation What if A and B are both at 56Mbps, and C is often at 2Mbps? Slowest node gets the most absolute time on channel? B Timeshare A B C C A Throughput Fairness vs Temporal Fairness Opportunistic Scheduling Goal Exploit short-time-scale channel quality variations to increase throughput. Issue Maintaining temporal fairness (time share) of each node. Challenge Channel info available only upon transmission Coherence Interval The time duration over which a channel is statistically likely to remain stable. This interval ranges from (122ms) - (5ms) based on node motion at speeds of (1 m/s) - (20 m/s). OAR was designed such that transmissions do not exceed the coherence interval “most” of the time. Coherence Interval OAR Transmission Opportunistic Auto Rate (OAR) Poor connections transmit one data packet per RTS/CTS connection. Good connections, hence faster rate, transmit multiple data packets. But maintain temporal fairness between good & bad connections by balancing the time using channel, not the number of packets. – i.e. (1 packet@2Mbps ~= 5 fast packets@11Mbps) OAR: Higher overall throughput, while maintaining temporal fairness properties of single rate IEEE 802.11 OAR Protocol Channel Condition Protocol BAD MEDIUM GOOD Pkts Rate Pkts Rate Pkts Rate 802.11 1 2 1 2 1 2 802.11b 1 2 1 5.5 1 11 OAR 1 2 3 5.5 5 11 Rates in IEEE 802.11b: 2, 5.5, and 11 Mbps Tx Rate Number of packets transmitted by OAR ~ Base Rate OAR Protocol (RBAR Based) Review: Receiver Based AutoRate (RBAR) [Bahl’01] Receiver controls the sender’s transmission rate Control messages sent at Base Rate ACK DATA CTS RTS destination source OAR Protocol (Multi-packet) Channel Condition Protocol BAD MEDIUM GOOD ACK Pkts Rate Pkts Rate Pkts Rate 802.11 1 2 1 2 1 2 802.11b 1 2 1 5.5 1 11 OAR 1 2 3 5.5 5 11 DATA ACK OAR - Opportunistic Auto Rate DATA ACK DATA CTS Once access granted, it is possible to send multiple packets if the channel is good RTS destination source Performance Comparison IEEE 802.11 Transmitter Receiver R D1 C A Observation II RBAR The total Transmitter R time by OARRis D1 in contention R D2 D3 approximately equal to total time spent in Receiver C by single-rate A C IEEE802.11 A Cfor an contention experiment spanning T seconds A OAR Observation I spent in by RBAR is R D1contention D2 per packet D3 exactly equal to the average time per packet spent Receiver C A IEEE802.11 A in contention for A single-rate Transmitter Time MAC Access Delay Simulation Back to back packets in OAR decrease the average access delay Increase variance in time to access channel Figure – On the left is 2Mbps – On the right is 5.5 Mbps Simulations Three Simulation experiments 1. Fully connected networks: all nodes in radio range of each other Number of Nodes, channel condition, mobility, node location 2. Asymmetric topology 3. Random topologies Implemented OAR and RBAR in ns-2 with extension of Ricean fading model [Punnoose et al ‘00] #1 Fully Connected Setup Every node can communicate with everyone Each node’s traffic is at a constant rate and continuously backlogged Channel quality is varied dynamically #1 Fully Connected Throughput Results OAR has 42% to 56% gain over RBAR Increase in gain as number of flows increases Note that both RBAR and OAR are significantly better than standard 802.11 (230% and 398% respectively) Variation in line of sight (K), mobility, and location distribution throughput all showed improvements with OAR. #2 Asymmetric Topology Setup Low speed (L) High Speed (H) A B Asymmetric topology simulated above in 4 different combinations of channel conditions – A and B are simulated at slow (2Mbps) and fast (11Mbps) – Each combination of slow/fast i.e. LL, HL, LH, HH compared between A & B concurrently communicating Sender of Flow B hears A and knows when to contend for channel, but sender in A has to discover a time slot #2 Asymmetric Topology Results OAR maintains time shares of IEEE 802.11 Significant gain over RBAR #3 Random Topologies Setup A pair are moved across a communication range Nodes are uniformly distributed over area similar to test setup #1 #3 Random Topologies Results Gains are similar as before despite changes Throughput is 40-50% improved as compared to RBAR despite motion of a node pair. Integration with IEEE 802.11 Options to hold the channel and send multiple packets – Fragmentation* A mechanism in IEEE 802.11 to send multiple frames Each frame/ACK acts as virtual RTS/CTS Use of more-fragment-flag in Data packets – Contention window set to zero – Packet bursting (802.11e) Transmit as many frames as you like up to threshold *Method used in study Discussion Issues Not enough packets to fill a slot – If running at “Good” 11Mbps with 5 packets allowed, but only have 2 packets to send. Then other nodes NAV tables are wrong (silent for 5 instead of 2). Authors Fix: “More Fragments” indicator in the data packet. Upon hearing, nodes revert to RBAR. Problem: Hidden terminals would still have incorrect NAV tables, and would remain silent longer than needed. (Unless the data ACK has a “More Fragments ACK.”) Discussion Issues Channel condition changes during multi-packet transmission. – Channel gets worse Later packets get corrupted – Channel gets better Wasted channel capacity waiting for packets to finish – Authors propose adding RSH messages to notify receiver of these updates and adapt the rate. The RSH is in the header of the data packet, and would allow changing speed mid transmission. Discussion Issues Ad Hoc Networks considerations – Needed more variety in the network topology. Fully connected isn’t very interesting in Ad Hoc Networks – Data traffic patterns. I.e. short bursts of traffic vs continuous traffic. – No power considerations studied or mentioned Discussion Issues Increase variance in time to access channel – Real-time traffic (like voice) is impacted. Sometimes there would be more delay before you hear “something.” – Short term fairness gets worse! – Trade throughput for a higher worst case time to access channel