The evolution of Internet of Things (IoT) has given a new direction to our day to day activities. As IoT strongly relies on sensors of several kinds, a large amount of sensory data is generated daily. The vast heterogeneity among the generated data items induces new challenges. Moreover, Quality of Service (\mathrm {Q}\mathrm {o}\mathrm {S}) requirements vary widely based on the application versatility. Traditionally, the processing and analysis of these data are performed at a distant server. However, off-site data processing tends to waste a lot of system resources and incurs high overhead with low QoS for the running applications. This brings in the need of low-cost in-network data management techniques that also suits the resource constrained nature of the sensor networks. This work aims to study the need of context adaptive sensing coupled with invocation of in-network data fusion strategy in a smart building environment. Based on the planned testbed we also justify the effectiveness of the proposed system by considering contextual parameters and in-network data fusion using Kalman Filter. © 2019 IEEE.