On the Scalability of Path Exploration Using Opportunistic Path-Vector Routing

Abstract

Abstract-Today various highly-efficient protocols exist as wellknown solutions to a wide-range of routing problems. However, there exist a large variety of scenarios where routing protocols do not perform well due to peculiar nature of these routing problems (e.g. dynamism of highly mobile ad-hoc networks, large scale of social, p2p networks, QoS-aware Inter-domain routing). These routing problems can be drawn from various research areas like wireless networks, social networks, and even Physics through fluid dynamics. In this work, we analyze opportunistic path-vector protocols as a solution to some of these mentioned problems. Our approach is to redefine routing problem as a set of smaller scale problems which can be solved locally without requiring a global coordination but local communication. We also provide guidelines on how to solve these redefinedlocalized routing problems efficiently. Our analysis show that our method provide a good compromise between scalability and opportunity through smartly randomized (non-deterministic) choices. Although we analyze this tradeoff for specific cases in this work, a wider range of routing problems can be actually targeted through transformation methods with reasonable-costs.

Path-Vector Contract Routing

“…With the increase in BTTL and MAXFWD parameter values, PVCR becomes more capable of discovering shorter paths due to increase in path inquiry packet limits as can be seen in Figure 6. We have not observed any performance degradation for path stretch in compared to our previous study [12] due to forwarding policies. Path stretch values between 1.5 and 2 indicate that PVCR is not only able to find great number of paths but also paths with high quality.…”

“…These results indicate significant improvements over our previous work which lacks ISP forwarding policies, i.e., BTTL distribution according to neighbor centrality. In fact, PVCR is able to perform 62% path exploration success for LP50 scenario given same BTTL and MAXFWD parameter values in our previous study [12].…”

“…Although these signaling protocols enable ISPs to negotiate construction of end-to-end paths with technical and economic properties, they delegate exploration of such AS-level paths to higher layers. Exploration of these end-to-end paths can be challenging [11], [12], as authors in [11] have proposed several heuristicbased solutions to the AS-level path exploration problem. However, these pre-computation based methods still incur high computation costs as their scalability will become more problematic with the flattening of the Internet and increasing connectivity among ISPs.…”

“…PATH-VECTOR CONTRACT ROUTING Simply demonstrated by Figure 2, PVCR path exploration process starts with a path inquiry initiated by a source ISP. Path inquiry packet describes the properties (e.g.,QoS, price and duration) of an end-to-end (e2e) path whose composition [12] is desired between a source and a destination ISP. Through forwarding path inquiries, ISPs along the way either choose to participate in composing such an end-to-end path or returning the offer by dropping (filter) the inquiry packet.…”