The proliferation of sensors as part of the Internet-of-Things (IoT) paradigm envisions a world of interconnected, interdependent deployments, involving one or multiple radio technologies. Multiradio mesh deployments facilitated using a network of low-cost IoT platforms, such as Raspberry Pis (RPis), ASUS Tinker Board S, Banana Pi M64, etc., seem an ideal fit as wireless backbones for such deployments, with many of these platforms having built-in support for multiple wireless technologies. However, the state-of-The-Art routing solutions need to be revisited to suit the capabilities of low-power multiradio solutions, especially while performing best effort transmissions. This is because a majority of the existing routing solutions favor channel diversity while routing, ignoring the possible overheads associated with protocol conversions. We argue that in reality, these overheads can become quite significant in a multiradio network where the radios are quite different in terms of packet formats, sizes, data rates, etc. To prove this, we estimate the impact of these overheads using an experimental study performed across three popular IoT technologies-WiFi, Bluetooth, and ZigBee, using a real-Time RPi-based testbed. We stress on how severe this impact could be, by estimating how often these conversions happen in a randomly chosen multihop path, using a probabilistic mathematical framework. Based on this, we develop Chorus, a first-of-its-kind routing algorithm, which estimates the least overhead path, inclusive of conversion overheads. Chorus achieves this by developing a comprehensive layered graph modeling of the given multiradio mesh network. The modeling is capable of handling all conversion overheads, including the factors associated with the possible fragmentation and aggregation. We implement Chorus in an indoor testbed comprising of RPis and demonstrate the performance improvement while routing across the network of RPis. Our results estimate that Chorus can provide an improvement close to 33% in battery usage per RPi with an average reduction of 25% in packet drops. © 2014 IEEE.