Modeling and analyzing the correctness of geographic face routing under realistic conditions

Geographic protocols are very promising for wireless ad hoc and sensor networks due to the low state storage and low message overhead. Under certain idealized conditions, geographic routing – using a combination of greedy forwarding and face routing – has been shown to work correctly and efficiently. In this work we model and analyze the correctness of geographic routing under non-ideal realistic conditions. We present a systematic methodology for micro-level behavioral analysis that shows that conditions that violate the unit-graph assumption of network connectivity, such as location errors, obstacles and radio irregularity, cause failure in planarization and consequently face routing. We then discuss the limitations of fixing these failures and prove that local algorithms that use only information up to a limited number of hops are not sufficient to guarantee face routing delivery under arbitrary connectivity. In addition, we analyze the effect of location errors in more detail to identify the possible protocol error scenarios and their conditions. We present results from an extensive simulation study about the effects of location errors on GPSR and GHT to quantify their performance degradation at different error ranges, distributions and error models. Based on our analysis we present a potential fix based on local information sharing that improves the performance significantly but does not remove all failures. Finally, we conclude that in order to avoid all failures under arbitrary connectivity, we need a non-local algorithm that can search or propagate information for an unlimited number of hops.