1) The document proposes CANS, a Congestion-Adaptive and Small stretch Emergency Navigation algorithm using wireless sensor networks (WSNs) to guide people to safely exit dangerous areas during emergencies.
2) CANS uses a level set method to track the evolution of hazardous area boundaries and the exit over time. This allows people near hazards to be routed on slightly longer but less congested paths, while those further away use shorter paths.
3) CANS builds a "hazard map" by assigning sensor nodes near hazards to groups with different risk levels. This scatters people across paths in different groups to reduce congestion. It can also adapt to changing hazards over time.