This paper presents our design and experience with a microkernel-inspired approach to host networking called Snap. Snap is a userspace networking system that supports Google's rapidly evolving needs with flexible modules that implement a range of network functions, including edge packet switching, virtualization for our cloud platform, traffic shaping policy enforcement, and a high-performance reliable messaging and RDMA-like service. Snap has been running in production for over three years, supporting the extensible communication needs of several large and critical systems.

The remarkable accessibility of modern flying robots makes them an attractive platform for environmental sensing. However, low cost and ease of use are currently incompatible with large payloads, severely limiting the choice of sensor and ultimately modality. This paper describes the design of a system for using a small infrared thermometer to estimate the surface temperature over an area that is large compared to the area measured by the sensor, by mounting it on a flying robot. We leverage a priori knowledge about the spatial statistics of the phenomena under measure in order to plan an informative sampling path, fusing observations by Gaussian process regression. Our approach is designed to be evaluated in an indoor testbed, in which a quadrotor, in cooperation with simulated static sensing nodes, estimates the spatial distribution of surface temperature over a controlled thermal gradient. We perform extensive systematic experimentation both in simulation and our real-world testbed environment, with our algorithm estimating surface temperature to an accuracy of up to 2.1 °C over a 16 m2 area ranging in value from 25-65 °C.

Environmental processes are often severely oversampled. As sensor networks become more ubiquitous for this purpose, increasing network longevity becomes ever more important. Radio transceivers in particular are a great source of energy consumption, and many networking algorithms have been proposed that seek to minimize their use. Traditionally, such approaches are often data agnostic, i.e., their performance is not dependent on the properties of the data they transport. In this paper we explore algorithms that exploit environmental relationships in order to reduce the amount of transmitted data while maintaining expected levels of accuracy. We employ a realistic testing environment for evaluating the power savings brought by such algorithms, based on Sensorscope, a commercial sensor network product for environmental monitoring. We implement and test a suppression-based data collection algorithm from literature that to our knowledge has never been implemented on a real system, and propose modifications that make it more suitable for real-world conditions. Using a custom extension board developed for in situ power monitoring, we show that while the algorithms greatly reduce the amount of energy spent on transmitting packets, they have no effect on the real system's overall power consumption due to its preexisting network architecture.

We propose two contrasting approaches to the scalable distributed control of a swarm of self-assembling miniaturized robots, specifically the formation of chains of a desired length: (1) a deterministic controller in which robots communicate with each other in order to directly limit the size of each chain, and (2) a probabilistic controller where the average chain size is controlled by the probability a robot will choose to leave its chain. We demonstrate the feasibility of both approaches by implementing them on a real swarm of Alice robots. Using Webots, a realistic simulator for mobile robotics, and macroscopic models based on the Chemical Reaction Network (CRN) framework, we investigate the limitations of the deterministic controller and demonstrate the existence of optimal parameters for the probabilistic controller where exploration and exploitation are well balanced, thus favoring the formation of larger chains.