12.07.2015 Views

Wireless Ad Hoc and Sensor Networks

Wireless Ad Hoc and Sensor Networks

Wireless Ad Hoc and Sensor Networks

SHOW MORE
SHOW LESS
  • No tags were found...

You also want an ePaper? Increase the reach of your titles

YUMPU automatically turns print PDFs into web optimized ePapers that Google loves.

Predictive Congestion Control for <strong>Wireless</strong> <strong>Sensor</strong> <strong>Networks</strong> 433Then, the DPC protocol is briefly described, <strong>and</strong> the used metrics arehighlighted.9.2.1 Congestion Prediction <strong>and</strong> MitigationTo predict the onset of congestion, the proposed scheme uses both queueutilization <strong>and</strong> the transmission power under the current channel state ateach node. When nodes in a network become congested, the traffic willaccumulate at the nodes because there will be an excess amount of incomingtraffic over the outgoing one. Hence, the queue utilization has beenselected as an indicator of the onset of congestion.On the other h<strong>and</strong>, in wireless networks during fading, the availableb<strong>and</strong>width is reduced <strong>and</strong> the outgoing rate will be lowered. Consequently,input <strong>and</strong> output buffers will accumulate the incoming trafficindicating the onset of congestion. The channel fading is estimated byusing the feedback information provided by the DPC protocol (Zawodniok<strong>and</strong> Jagannathan 2004) for the next packet transmission. The DPCalgorithm predicts the channel state for the subsequent time interval <strong>and</strong>calculates the required power. If this power exceeds the maximum threshold,then the channel is considered to be unsuitable for transmission, <strong>and</strong>the proposed congestion control scheme can initiate the backoff processby reducing incoming traffic. Hence, information from DPC can be utilizedto predict the onset of congestion due to fading channels.Once the onset of congestion has been assessed, different strategies canbe applied to prevent it from occurring. We propose a scheme, with a goalto prevent <strong>and</strong> minimize the effect of congestion while ensuring weightedfairness. When applied at every node in the wireless sensor network, itwill render a fair <strong>and</strong> distributed congestion control scheme. Theemployed algorithms minimize queue overflows at a particular node byregulating the incoming flow. The admissible incoming traffic is calculatedbased on three factors:Predicted outgoing flow — The outgoing flow is periodically measured,<strong>and</strong> an adaptive scheme is used to accurately predict theoutgoing flow in the next period; moreover, the next hop nodecan reduce the outgoing flow assessment by applying a controlover its incoming flows.<strong>Wireless</strong> link state — The predicted outgoing flow rate is furtherreduced when the DPC protocol predicts a severe channel fading,which will disrupt communication on the link.Queue utilization — The algorithm restricts the incoming flow basedon the current queue utilizations <strong>and</strong> predicted outgoing flow,thus reducing buffer overflows.

Hooray! Your file is uploaded and ready to be published.

Saved successfully!

Ooh no, something went wrong!