Dynamic Subscription Permission: Extending the Depth of Demand-Controlled Flooding

Abstract

We present a method for improving publication delivery range in self-balancing supply/demand (SBSD) routing. This method, dynamic subscription permission (DSP), restricts each subscription type to an estimated connected dominating set and dynamically restricts routing to adapt to local density variations. Given a theoretical mobile device population n 0 comprising one connected dominating set and an actual network population n > n 0 , SBSD using DSP increases subscription propagation areas by up to a factor of (n/n 0 ). Results are presented for road-based mobility models.

Demand-driven publish/subscribe in mobile environments

“…Indeed, if the flooding depth estimates are exactly correct, the subscription will not reach a more distant publisher and so not find a match. Although SBSD is not well-suited for routing over long distances, in high-mobility environments the performance of route-building protocols can be rather worse, as was observed in [30].…”

“…The functions used in [29] and [30], both of which used a different scheduling and queuing scheme, also differ from the one used here. Our utility function not only has theoretically desirable attributes but also has been refined to improve publication delivery and response time.…”

“…This paper introduces three significant improvements over the earlier SBSD models used in [29,30]. First, there is no longer an arbitrary limit to the number of subscription replicas stored at each broker; brokers retain all unexpired replicas for possible future broadcast.…”

“…Our MAC here is the same extension of R-TDMA [28] used in [29,30] and which was termed 4H. R-TDMA allocates each node an equal share of the broadcast time within its two-hop neighborhood.…”

“…We believe these attributes make SBSD a promising approach for applications in potentially very large MANETs, such as real-time transportation information delivery. SBSD's performance as a VANET application was presented in [30]. Our focus here is on the properties of SBSD, so we test SBSD under random mobility, which is more tractable for equation-based analysis.…”