The lack of a sustainable power source to substitute batteries for long-term applications limits the widespread deployment of wireless sensor nodes in this era of the Internet of Things. Conventional linear Vibration Energy Harvesters are inefficient in converting ambient mechanical energy into usable electrical energy owing to their narrow frequency bandwidth when harnessing mechanical energy that is spread over a wide range of frequencies. In this work, we design, develop and demonstrate high power density nonlinear wideband energy harvesters using novel tapered spring architectures in an autonomous wireless sensor node system. These spring structures exhibit a nonlinear restoring force arising from the atypical stress distribution that can be additionally tuned by changing the taper-ratio in the structure. We investigate different tapering designs in order to achieve optimal spring hardening nonlinearities. This nonlinearity aids in widening the operable bandwidth, making the harvesters suitable for scavenging energy from real-world broadband vibrations. We obtain power densities of the order of 2660 µW/cm3g2 in the nonlinear energy harvester, outpacing most contemporary energy scavengers. We present a modified Perturb and Observe algorithm that allows tracing of the maximum power point in the context of non-stationary vibration conditions. We use the fabricated nonlinear device to power a wireless sensor node that reports on vital physical parameters (humidity, temperature), thereby enabling a resilient remote data acquisition system. This demonstrates the potential of our design to provide a sustainable energy source for platforms within the Internet of Things.