13.07.2015 Views

Page 2 Lecture Notes in Computer Science 2865 Edited by G. Goos ...

Page 2 Lecture Notes in Computer Science 2865 Edited by G. Goos ...

Page 2 Lecture Notes in Computer Science 2865 Edited by G. Goos ...

SHOW MORE
SHOW LESS

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

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

62 Y. Ge, T. Kunz, and L. LamontFig. 1. Simple network. An edge between two nodes <strong>in</strong>dicates that the two nodes connected <strong>by</strong>this edge are with<strong>in</strong> reach of each other. The edge weight represents the QoS l<strong>in</strong>k attribute weare <strong>in</strong>terested <strong>in</strong>, available bandwidth.OLSR is a rout<strong>in</strong>g protocol for best-effort traffic, with emphasis on how to reducethe overhead. So <strong>in</strong> its MPR selection, the node selects the neighbor that covers themost unreachable 2-hop neighbors as MPR. However, <strong>in</strong> QoS rout<strong>in</strong>g, <strong>by</strong> such MPRselection mechanism, the “good quality” l<strong>in</strong>ks may be “hidden” to other nodes <strong>in</strong> thenetwork. As an example, we will consider the network topology <strong>in</strong> Fig. 1. In theOLSR MPR selection algorithm, node B will select C as its MPR. So for all the othernodes, they only know that they can reach B via C. Obviously, when D is build<strong>in</strong>g itsrout<strong>in</strong>g table, for dest<strong>in</strong>ation B, it will select the route D-C-B, whose bottleneckbandwidth is 3, the worst among all the possible routes. Also, when “bandwidth” isthe QoS constra<strong>in</strong>t, nodes can no longer use the “shortest hops path” algorithm asproposed <strong>in</strong> OLSR. Because of these limitations of OLSR <strong>in</strong> QoS rout<strong>in</strong>g, the QoSOLSR version revises it <strong>in</strong> two aspects: MPR selection and rout<strong>in</strong>g table computation.The decision on how each node selects its MPRs is essential to determ<strong>in</strong><strong>in</strong>g the optimalbandwidth route <strong>in</strong> the network. In select<strong>in</strong>g the MPRs, a “good bandwidth” l<strong>in</strong>kshould not be omitted. Based on this idea, we previously explored three revised MPRselection algorithms [5]. In this paper, we implement the best variant (OLSR_R2) <strong>in</strong>OPNET to compare its performance with the orig<strong>in</strong>al OLSR protocol. The idea beh<strong>in</strong>dOLSR_R2 is to select the best bandwidth neighbors as MPRs until all the 2-hopneighbors are covered.Table 2. Node B’s MPR(s), us<strong>in</strong>g OLSR_R2.Node 1 Hop Neighbors 2 Hop Neighbors MPR(s)B A, C, F, G D, E A, FAmong node B’s neighbors, A, C, and F have a connection to its 2-hop neighbors.Among them, l<strong>in</strong>k BA has the highest bandwidth. So A is first selected as B’s MPR,and the 2-hop neighbor D is covered. Similarly, F is selected as MPR next and E iscovered, so all 2-hop neighbors are covered and the algorithm term<strong>in</strong>ates. This revisedOLSR MPR selection algorithm improves the chance that a better bandwidthroute is found. However, <strong>by</strong> us<strong>in</strong>g such algorithm, the overhead also <strong>in</strong>creases becausethe number of MPRs <strong>in</strong> the network is <strong>in</strong>creased.

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

Saved successfully!

Ooh no, something went wrong!