From Real Neighbors to Imaginary Destination: Emulation of Large Scale Wireless Sensor Networks

Abstract

The ultimate test for many network layer protocols designed for wireless sensor networks would be to run on a large scale testbed. However, setting up a real-world large scale wireless sensor network (WSN) testbed requires access to a huge surface as well as extensive financial and human resources. Due to limited access to such infrastructures, the vast majority of existing theoretical and simulation studies in WSN are far from being validated in realistic environments. A more affordable approach is needed to provide preliminary insights on network protocol performances in large WSN. To replace large and expensive realistic testbeds, we introduce a novel approach to emulation. We propose a specifically designed experimental setup using a relatively small number of nodes forming a real one-hop neighborhood used to emulate any real WSN. The source node is a fixed sensor, and all other sensors are candidate forwarding neighbors towards a virtual destination. The source node achieves one forwarding step, then the virtual destination position and neighborhood are adjusted. The same source is used again to repeat the process. The main novelty is to spread available nodes regularly following a hexagonal pattern around the central node, used as the source, and selectively use subsets of the surrounding nodes at each step of the routing process to provide the desired density and achieve changes in configurations. Compared to real testbeds, our proposition has the advantages of emulating networks with any desired node distribution and densities, which may not be possible in a small scale implementation, and of unbounded scalability since we can emulate networks with an arbitrary number of nodes. Finally, our approach can emulate networks of various shapes, possibly with holes and obstacles. It can also emulate recovery mode in geographic routing, which appears impossible with any existing approach.