Energy harvesting is increasingly used as a long-term energy supply for the Internet of Things, wireless sensor networks, and cyber-physical systems. However, the challenge of mitigating the variability of energy harvesting sources needs to be addressed before ubiquitous adoption can happen. Otherwise, an unreliable operation of devices with frequent shutdowns during times of energy scarcity would be encountered. One finds probabilistic performance metrics helpful in designing power management solutions for the long-term operation of energy harvesting nodes. These metrics include the probability of battery depletion, expected energy consumption, expected battery level, and similar. This article proposes a stochastic modeling technique and corresponding analysis which can provide such metrics. Our advanced analysis is based on Markov chains. By modeling harvested energy with random variables, new and existing energy management policies are analyzed and compared. We propose an adaptive energy management strategy inspired by mixed-criticality systems. In the proposed strategy, the system can degrade or drop less essential tasks in real time to ensure a graceful degradation of service in adverse harvesting conditions. We compare the proposed energy management approach to several existing alternatives. To this end, we conduct extensive simulations for indoor and outdoor environments, where our strategy matches or outperforms the state of the art. Initial results also validate the high precision of our stochastic model and analysis in comparison to simulations using real-world data.

Abstract Mixed-criticality systems often need to fulfill safety standards that dictate different requirements for each criticality level, for example given in the ‘probability of failure per hour’ format. A recent trend suggests designing this kind of systems by jointly scheduling tasks of different criticality levels on a shared platform. When this is done, the usual assumption is that tasks of lower criticality are degraded when a higher criticality task needs more resources, for example when it overruns a bound on its execution time. However, a way to quantify the impact this degradation has on the overall system is not well understood. Meanwhile, to improve schedulability and to avoid over-provisioning of resources due to overly pessimistic worst-case execution time estimates of higher criticality tasks, a new paradigm emerged where task’s execution times are modeled with random variables. In this paper, we analyze a system with probabilistic execution times, and propose metrics that are inspired by safety standards. Among these metrics are the probability of deadline miss per hour, the expected time before degradation happens, and the duration of the degradation. We argue that these quantities provide a holistic view of the system’s operation and schedulability.

In this work, we present a formal study on optimizing the energy consumption of energy harvesting embedded systems. To deal with the uncertainty inherent in solar energy harvesting systems, we propose the Stochastic Power Management (SPM) scheme, which builds statistical models of harvested energy based on historical data. The proposed stochastic scheme maximizes the lowest energy consumption across all time intervals while giving strict probabilistic guarantees on not encountering battery depletion. For situations where historical data is not available, we propose the use of (i) a Finite Horizon Control (FHC) scheme and (ii) a non-uniformly scaled energy estimator based on an astronomical model, which is used by FHC. Under certain realistic assumptions, the FHC scheme can provide guarantees on minimum energy usage that can be supported over all times. We further propose and evaluate a piece-wise linear approximation of FHC for efficient implementation in resource-constrained embedded systems. With extensive experimental evaluation for eight publicly available datasets and two datasets collected with our own deployments, we quantitatively establish that the proposed solutions are highly effective at providing a guaranteed minimum service level and significantly outperform existing solutions.

Micro aerial vehicles (MAVs) provide data such as images and videos from an aerial perspective, with data typically transferred to the ground. To establish connectivity in larger areas, a fleet of MAVs may set up an ad-hoc wireless network. Packet forwarding in aerial networks is challenged by unstable link quality and intermittent connectivity caused by MAV movement. We show that signal obstruction by the MAV frame can be alleviated by adapting the MAV platform, even for low-priced MAVs, and the aerial link can be properly characterized by its geographical distance. Based on this link characterization and making use of GPS and inertial sensors on-board of MAVs, we design and implement a motion-driven packet forwarding algorithm. The algorithm unites location-aware end-to-end routing and delay-tolerant forwarding, extended by two predictive heuristics. Given the current location, speed, and orientation of the MAVs, future locations are estimated and used to refine packet forwarding decisions. We study the forwarding algorithm in a field measurement campaign with quadcopters connected over Wi-Fi IEEE 802.11n, complemented by simulation. Our analysis confirms that the proposed algorithm masters intermittent connectivity well, but also discloses inefficiencies of location-aware forwarding. By anticipating motion, such inefficiencies can be counteracted and the forwarding performance can be improved.