Wireless Ad Hoc and Sensor Networks
Wireless Ad Hoc and Sensor Networks Wireless Ad Hoc and Sensor Networks
Distributed Fair Scheduling in Wireless Ad Hoc and Sensor Networks 345the next subsection. Next, when the packet is received at an intermediatenode, the packet’s weight is updated and queued accordingly.7.6.2.2 Protocol ImplementationTo achieve fairness at the scheduling level, the proposed ADFS protocolimplements the SFQ scheme, defined as follows:1. On arrival, a packet of flow f , is stamped with start tag ( ) ,defined asp fj{ } ≥( f )= ( ( f )) ( f )Sp j vAp j Fp j−1max , , j 1,jwhere Fp ( f ), finish tag of packet p j jf , is defined as Fp ( f )=Sp ( jf ) jl+f , j≥1, φ f0where Fp ( f ) = 0 and φ f is the weight of flow f .2. Initially, the virtual time of the sensor node is set to zero. Duringtransmission, the node’s virtual time at time t , vt () is defined tobe equal to the start tag of the packet being transmitted at time t.At the end of a transmission, vt () is set to the maximum of finishtag assigned to any packets that have been transmitted by time t.3. Packets are transmitted in the increasing order of the start tags;ties are broken arbitrarily.7.6.2.3 Dynamic Weight Adaptation and Backoff Interval CalculationTo account for the changing traffic and channel conditions that affect thefairness and end-to-end delay, the weights for the flows are updateddynamically. The ADFS, which is detailed in Section 7.1 to Section 7.4,calculates the backoff interval by using the packet weight and updatedat each node due to weight adaptation.7.6.2.4 Distributed Power Control and ResetThe basic communication used to send a user packet requires a four-wayhandshake exchange which consists of four frames:1. Request-to-send (RTS) — from source to destination2. Clear-to-send (CTS) — from destination to source3. DATA frame — from source to destination4. Acknowledgment (ACK) — from destination to sourceSp fjAll these frames are transmitted over the single radio channel. Hence, thecommunication between the nodes is carried over shared half duplex medium.
346 Wireless Ad Hoc and Sensor NetworksAdditionally, handshake between other nodes can exist at the same timeinstance. As a result, the interference can increase due to multiple nodescommunicating with their destinations by accessing the channel. The estimatedtransmitter power has to overcome such an interference, which isnormally not known. In addition, the packet transmission as well as arrivaltimes will vary between frames due to the following reasons. First, theframes within the 4-way handshake have different size, ranging from fewbytes (ACK) to over 2500 bytes (DATA) due to the frame type. Second, thedelay between two consecutive handshakes can vary because of channelcontention. As the result, the estimation error will also vary depending onthese time gaps. Therefore, the target signal-to-interference ratio (SIR) hasto be chosen to overcome the worst-case scenario due to these uncertainties.Thus in our implementation, the target SIR value is calculated by multiplyingthe minimum SIR with a safety factor.In the real scenario, it is possible that channel conditions can changetoo quickly, which will prevent any algorithm to accurately estimate thepower value. Then, the frame losses will occur. Two mechanisms areintroduced to overcome or mitigate such problems: increase of transmissionpower in the case of retransmission and reset of power in the caseof a long idle connection.A retransmission indicates that the received signal has been attenuated.A simple approach would be to retransmit the frame with the same poweras the first transmission by hoping that the channel conditions are better.However, the channel attenuation or interferences could also be worse thanbefore. To overcome this problem, an active approach can be used. In thisapproach, for each retransmission, the transmission power is increased by acertain safety margin. Unfortunately, this will increase interference as wellas power consumption. However, experiments indicated that active adaptationof power with the DPC yielded higher throughput and resulted in lessretransmissions when compared to the passive method where no safety marginwas utilized. As the result, safety margin is used with the DPC scheme.Additionally, the estimated power value for a given destination canbecome inaccurate when significant delays are present between any twoconsecutive frames or delayed feedback. To overcome this, after certainidle interval, the DPC algorithm will reset the transmission power to themaximum value defined for the network. The DPC process described inChapter 6 is then restarted from the beginning.7.6.3 Energy-Aware MAC ProtocolFor this DPC implementation, the original MAC protocol for 802.11 hasbeen modified. These modifications occur at different layers. Furthermore,an idea, which is presented in Zawodniok and Jagannathan (2004), for adhoc networks, is used in the energy-aware protocol.
- Page 317 and 318: 294 Wireless Ad Hoc and Sensor Netw
- Page 319 and 320: 296 Wireless Ad Hoc and Sensor Netw
- Page 321 and 322: 298 Wireless Ad Hoc and Sensor Netw
- Page 323 and 324: 300 Wireless Ad Hoc and Sensor Netw
- Page 326 and 327: 7Distributed Fair Scheduling in Wir
- Page 328 and 329: Distributed Fair Scheduling in Wire
- Page 330 and 331: Distributed Fair Scheduling in Wire
- Page 332 and 333: Distributed Fair Scheduling in Wire
- Page 334 and 335: Distributed Fair Scheduling in Wire
- Page 336 and 337: Distributed Fair Scheduling in Wire
- Page 338 and 339: Distributed Fair Scheduling in Wire
- Page 340 and 341: Distributed Fair Scheduling in Wire
- Page 342 and 343: Distributed Fair Scheduling in Wire
- Page 344 and 345: Distributed Fair Scheduling in Wire
- Page 346 and 347: Distributed Fair Scheduling in Wire
- Page 348 and 349: Distributed Fair Scheduling in Wire
- Page 350 and 351: Distributed Fair Scheduling in Wire
- Page 352 and 353: Distributed Fair Scheduling in Wire
- Page 354 and 355: Distributed Fair Scheduling in Wire
- Page 356 and 357: Distributed Fair Scheduling in Wire
- Page 358 and 359: Distributed Fair Scheduling in Wire
- Page 360 and 361: Distributed Fair Scheduling in Wire
- Page 362 and 363: Distributed Fair Scheduling in Wire
- Page 364 and 365: Distributed Fair Scheduling in Wire
- Page 366 and 367: Distributed Fair Scheduling in Wire
- Page 370 and 371: Distributed Fair Scheduling in Wire
- Page 372 and 373: Distributed Fair Scheduling in Wire
- Page 374 and 375: Distributed Fair Scheduling in Wire
- Page 376 and 377: Distributed Fair Scheduling in Wire
- Page 378 and 379: Distributed Fair Scheduling in Wire
- Page 380 and 381: 8Optimized Energy and Delay-Based R
- Page 382 and 383: Optimized Energy and Delay-Based Ro
- Page 384 and 385: Optimized Energy and Delay-Based Ro
- Page 386 and 387: Optimized Energy and Delay-Based Ro
- Page 388 and 389: Optimized Energy and Delay-Based Ro
- Page 390 and 391: Optimized Energy and Delay-Based Ro
- Page 392 and 393: Optimized Energy and Delay-Based Ro
- Page 394 and 395: Optimized Energy and Delay-Based Ro
- Page 396 and 397: Optimized Energy and Delay-Based Ro
- Page 398 and 399: Optimized Energy and Delay-Based Ro
- Page 400 and 401: Optimized Energy and Delay-Based Ro
- Page 402 and 403: Optimized Energy and Delay-Based Ro
- Page 404 and 405: Optimized Energy and Delay-Based Ro
- Page 406 and 407: Optimized Energy and Delay-Based Ro
- Page 408 and 409: Optimized Energy and Delay-Based Ro
- Page 410 and 411: Optimized Energy and Delay-Based Ro
- Page 412 and 413: Optimized Energy and Delay-Based Ro
- Page 414 and 415: Optimized Energy and Delay-Based Ro
- Page 416 and 417: Optimized Energy and Delay-Based Ro
Distributed Fair Scheduling in <strong>Wireless</strong> <strong>Ad</strong> <strong>Hoc</strong> <strong>and</strong> <strong>Sensor</strong> <strong>Networks</strong> 345the next subsection. Next, when the packet is received at an intermediatenode, the packet’s weight is updated <strong>and</strong> queued accordingly.7.6.2.2 Protocol ImplementationTo achieve fairness at the scheduling level, the proposed ADFS protocolimplements the SFQ scheme, defined as follows:1. On arrival, a packet of flow f , is stamped with start tag ( ) ,defined asp fj{ } ≥( f )= ( ( f )) ( f )Sp j vAp j Fp j−1max , , j 1,jwhere Fp ( f ), finish tag of packet p j jf , is defined as Fp ( f )=Sp ( jf ) jl+f , j≥1, φ f0where Fp ( f ) = 0 <strong>and</strong> φ f is the weight of flow f .2. Initially, the virtual time of the sensor node is set to zero. Duringtransmission, the node’s virtual time at time t , vt () is defined tobe equal to the start tag of the packet being transmitted at time t.At the end of a transmission, vt () is set to the maximum of finishtag assigned to any packets that have been transmitted by time t.3. Packets are transmitted in the increasing order of the start tags;ties are broken arbitrarily.7.6.2.3 Dynamic Weight <strong>Ad</strong>aptation <strong>and</strong> Backoff Interval CalculationTo account for the changing traffic <strong>and</strong> channel conditions that affect thefairness <strong>and</strong> end-to-end delay, the weights for the flows are updateddynamically. The ADFS, which is detailed in Section 7.1 to Section 7.4,calculates the backoff interval by using the packet weight <strong>and</strong> updatedat each node due to weight adaptation.7.6.2.4 Distributed Power Control <strong>and</strong> ResetThe basic communication used to send a user packet requires a four-wayh<strong>and</strong>shake exchange which consists of four frames:1. Request-to-send (RTS) — from source to destination2. Clear-to-send (CTS) — from destination to source3. DATA frame — from source to destination4. Acknowledgment (ACK) — from destination to sourceSp fjAll these frames are transmitted over the single radio channel. Hence, thecommunication between the nodes is carried over shared half duplex medium.