An energy-efficient MAC protocol for infrastructure WLAN based on modified PCF/ DCF access schemes using a bidirectional data packet exchange Raúl Palacios, Fabrizio Granelli University of Trento Trento, Italy September 18th, 2012. Barcelona, Spain Danica Gajic, Andreas Foglar InnoRoute GmbH Munich, Germany UNIVERSITÀ DEGLI STUDI DI TRENTO
Outline Scenario IEEE 802.11 State of the art Problem Solution Bidirectional transmissions PCF/DCF Evaluation Summary Future work Result 2
A general picture The Base Station... communicates with mobile devices: coordinating transmissions, sending data and providing a wireless connection to the Internet. Internet Base Station E Pollution F CO2 Wi Fi S Lte UMTS Wireless channel EM waves You Tube THE PROBLEMS The energy consumption and carbon footprint of base are ever-increasing year after year. Mobile Data Traffic The radio interface severely limits the battery life of a mobile device. The mobile device communicates with the base station: monitoring the wireless channel, to transmit or receive data to or from the base station. 3
IEEE 802.11 MAC protocol Active mode Contention Free Period PCF Deterministic Polling Collision free Contention Period DCF Random Contention Collision avoidance Listen Transmit Receive Sleep Wake-up PS mode Z Z Z CFP Repetition Interval CFP Repetition Interval PIFS Contention Free Period Contention Period CFP CP Access Point B Point Coordination Function CE Distributed Coordination Function B PCF CE DCF 802.11 NAV NAV Time 4
Energy consumption in DCF Distributed Coordination Function (DCF) Source Destination Other DIFS CW RTS CTS Data NAV RTS NAV CTS NAV Data DIFS CW Time CTS RTS Data source control frames over-hearing Idle-listening collisions destination 5
Energy consumption in PCF Point Coordination Function (PCF) Access Point Polled PIFS B Poll+ D1 Contention Free Period +U1 +D2 +Poll Missing response PIFS Poll+ D3 + Null CE Other NAV Time Poll + D1 Polled? access point + D2 + Poll + U1 Poll + D3 + Null polling frames polling list variable data over-hearing idle-listening 6
Energy-efficient MAC solutions Remove control overhead Block acknowledgment and data aggregation (IEEE802.11e/n) Group-polling frames Alternate RTS/CTS with CTS polling, or negative CTS Reduce the number of (re)transmissions Transmission power control (TPC) RTS/CTS threshold, fragmentation threshold Physical-layer rate adaptation, alone and combined with TPC Minimize waiting time for transmitting data Redefine the binary exponential backoff (BEB) algorithm Sleep during backoff periods Sleep during the polling access and only wake up to send data Bidirectional transmissions: attach data to (IEEE802.11ad) 7
Our approach a destination, after a successful data reception, is not restricted to forward an acknowledgment to the source but can, instead, convey bidirectional data in exchange, with a packet of equal duration/shorter of the source packet Compatible with IEEE 802.11ad access mode of bidirectional transmissions 8
MAC enhancements for DCF Source Destination Other Source Destination Other DIFS DIFS CW CTS CTS Data Data in exchange Data NAV RTS NAV CTS NAV Data NAV RTS DIFS Defer access NAV CTS NAV Data CTS replaced by Data Data Data NAV RTS NAV CTS NAV Data DIFS DIFS Time New access (a) Standard DCF: a successful RTS/CTS handshake only allows an initiating source to send data to its intended destination. CW RTS RTS Time New access (b) Modified DCF: a successful RTS/CTS handshake can be used to convey bidirectional data between a source and its intended destination. CW RTS CW CW 9
MAC enhancements for PCF Access Point (AP) Polled Other reply with uplink data or a null frame. Contention Free Period Modified PCF Access Point (AP) Polled Other PIFS PIFS B B Poll + D1 D1 + U1 Poll replaced by D1 U1 Contention Free Period Point Coordination Function + Poll + D2 + D2 replaced by U1 Missing response NAV PIFS Missing response NAV PIFS D3 Poll + D3 + Null Poll Null CE CE Contention Period Distributed Coordination Function Dx: Frames sent from AP to station x Ux: Frames sent from station x to AP Reset NAV (a) Standard PCF: the access point sends downlink data and polls and each polled station must Null replaced by Time Contention Period Modified DCF Dx: Frames sent from AP to station x Ux: Frames sent from station x to AP Reset NAV (b) Modified PCF: each station, upon reception of downlink data, can send an uplink frame with equal duration of the downlink frame. Time 10
Simulation framework Access point - Energy contributor - Heavy traffic load - Continuous polling (PCF) Channel: - Errors - Collisions (DCF) - Empty slot (DCF) Stations - Energy contributors - Fixed traffic load - Arbitrary packet size 11
Simulation results: Throughput 45 40 11% Megabit/second 35 30 25 20 15 10 5 0 modified PCF standard PCF modified DCF standard DCF 0 5 10 15 20 25 30 35 40 45 No. of nodes Average system throughput with the number of. 30% PHY/MAC parameters (IEEE 802.11g) Payload (1500 bytes) PER (5%) Traffic (50 packets) 12
Simulation results: Energy effic. Joule/Megabit 4 3,5 3 2,5 2 1,5 1 modified PCF standard PCF modified DCF standard DCF 30% 11% 0,5 0 0 5 10 15 20 25 30 35 40 45 No. of nodes Average system energy consumption with the number of. Tx power (1.65 Watts) Rx power (1.4 Watts) Idle power (1.15 Watts) 13
Concluding remarks Proposed PCF/DCF MAC modifications: Higher throughput -> Better mobile services Less energy -> Longer battery life Feasible implementation into standardization Future work: new energy-saving strategies enabled by modified PCF/DCF protocols in infrastructure WLANs. Multiple rounds of bidirectional transmissions. Improved Power Save mode of PCF/DCF. Sleep during NAVs or sleep during most of time of a CFP. DCF can sleep in CFPs and PCF in CPs. 14
Thank you for your attention 15
Back-up: Simulation parameters Parameter Value Parameter Value 10μs Txpower 1.65Watt DIFS 28μs Rxpower 1.4Watt Tx. preamble 16μs Idlepower 1.15Watt Tx. PHY header 4μs Beacon/CF-end/ RTS 20bytes Slot duration 9μs Control Tx. rate 6Mbps Cwmin, CWmax 15, 1023 Null packet length 240bytes MAC header 34bytes Data payload 1500bytes Data Tx. Rate 48Mbps Packet error rate 5% No. of sta. 2-40 No. of packets 50 16
Back-up: Performance metrics System throughput (S) [Mbps] = [transmitted payload information]/ [slot time] SCRIPT C++ S = (Data_payload*N_tx)/total_time [Mbps] System Energy Efficiency (Eef) [Joule/bits] = [energy consumption]/[transmitted payload information] SCRIPT C++ Eef = E_consumption_total/ (Packet_length*N_tx) E_consumption_total = total (tx_time*tx_power + rx_time*rx_power + slot_time*idle_power) 17