Directional antennas are a promising solution for improving the range of client devices and capacity of wireless networks. Unfortunately, in LP-WAN systems directional antennas tend to be both large and expensive due to operating at subGHz frequencies. However, if a client device is willing to forgo improvements in antenna gain, it is possible to realize compact and low-cost antennas that provide spatial diversity control. In this paper, we show that by increasing spatial diversity in LPWAN clients with limited (or no) client gain, we can dramatically increase overall network capacity and improve client battery life by avoiding re-transmissions. This type of directional control can also be used for hot-spot management by more effectively load balancing clients across gateways.We performed a sensitivity analysis using a combination of real hardware and simulation to explore the impact of various switchable antenna geometries on network capacity under a variety of deployment configurations. We then designed and evaluated three prototype multi-sector array clients: (1) a switchable patch antenna configuration, (2) a digital phase-shift nulling configuration, and (3) a low-cost switched PCB element phase-shift system. Each design explores a different hardware cost vs antenna beam performance operating point. We experimentally see that our real antenna beam patterns, captured in an anechoic chamber, perform in a similar manner to our simulated prediction models in terms of beam pattern and in simulation improve network capacity by up to 28% from interference isolation alone and up to 95% when offloading hot spots between four gateways. We also perform a small measurement study of how often our final design changes its configuration when deployed over multiple days on a campus testbed.

Internet of Things (IoT) technologies have enabled numerous data-driven mobile applications and have the potential to significantly improve environmental monitoring and hazard warnings through the deployment of a network of IoT sensors. However, these IoT devices are often power-constrained and utilize wireless communication schemes with limited bandwidth. Such power constraints limit the amount of information each device can share across the network, while bandwidth limitations hinder sensors’ coordination of their transmissions. In this work, we formulate the communication planning problem of IoT sensors that track the state of the environment. We seek to optimize sensors’ decisions in collecting environmental data under stringent resource constraints. We propose a multi-agent reinforcement learning (MARL) method to find the optimal communication policies for each sensor that maximize the tracking accuracy subject to the power and bandwidth limitations. MARL learns and exploits the spatial-temporal correlation of the environmental data at each sensor’s location to reduce the redundant reports from the sensors. Experiments on wildfire spread with LoRA wireless network simulators show that our MARL method can learn to balance the need to collect enough data to predict wildfire spread with unknown bandwidth limitations.

Many Cyber-Physical Systems (CPS) have timing constraints that must be met by the cyber components (software and the network) to ensure safety. It is a tedious job to check if a CPS meets its timing requirement especially when it is distributed and the software and/or the underlying computing platforms are complex. Furthermore, the system design is brittle since a timing failure can still happen (e.g., network failure, soft error bit flip). In this article, we propose a new design methodology calledPlan Bwhere timing constraints of the CPS are monitored at runtime, and a proper backup routine is executed when a timing failure happens to ensure safety. We provide a model on how to express the desired timing behavior using a set of timing constructs in a C/C++ code and how to efficiently monitor them at the runtime. We showcase the effectiveness of our approach by conducting experiments on three case studies: (1) the full software stack for autonomous driving (Apollo), (2) a multi-agent system with 1/10th-scale model robots, and (3) a quadrotor for search and rescue application. We show that the system remains safe and stable even when intentional faults are injected to cause a timing failure. We also demonstrate that the system can achieve graceful degradation when a less extreme timing failure happens.

LoRa is a popular Low-Power Wide-Area Networking (LP-WAN) technology that allows devices powered by a ten year AA battery to connect to radio infrastructure miles away. One of the most promising features of LoRa is the ability to track the location of radios from a distance, enabling applications ranging from inventory tracking, smart infrastructure monitoring and structural health sensing. Yet, state-of-the-art LoRa localization systems experience errors of several tens or even hundreds of meters in location tracking, owing to the narrow bandwidth and limited battery life of LoRa devices.

Low-Power Wide-Area Networks (LP-WANs) are seeing wide-spread deployments connecting millions of sensors, each powered by a ten-year AA battery to radio infrastructure, often miles away. By design, iteratively querying all sensors in an LP-WAN may take several hours or even days, given the stringent battery limits of client radios. This precludes obtaining even an approximate real-time view of sensed information across LP-WAN devices over a large area, say in the event of a disaster, fault or simply for diagnostics.This paper presents QuAiL1, a system that provides a coarse aggregate view of sensed data across LP-WAN devices over a wide- area within a time span of just one LP-WAN packet. QuAiL achieves this by coordinating multiple LP-WAN radios to transmit their information synchronously in time and frequency despite their power constraints. We design each client’s transmission so that the base station can retrieve an approximate heatmap of sensed data by exploiting the spatial correlation of this data across clients. We further show how our system can be optimized for statistical and machine learning queries, all while maintaining the security and privacy of sensed data from individual clients. Our deployment over a 3 sq. km. LP-WAN deployment around CMU campus in Pittsburgh demonstrates a 4x faster information retrieval versus the state-of- the-art statistical methods to retrieve the spatial sensor heatmap at a desired resolution.

Recent years have witnessed novel designs of battery-free IoT tags using RF backscatter. Traditionally, they require a dedicated transmitter to excite the tag. However, such a deployment is infeasible at large scale. To counter this problem, researchers have proposed using ambient RF energy to power up the battery-free tag. In this poster, we evaluate if this ambient RF energy is sufficient to meet the power requirements of a battery-free tag in today's urban and rural areas. We also compare available ambient RF energy across different frequencies. Finally, we discuss open challenges in realising ambient backscatter systems in real-world.

Low-Power Wide Area Networks, such as LoRaWAN, are rapidly gaining popularity in the field of wireless sensing and actuation. While LoRaWAN is heavily studied in applications and performance, the concept of time has rarely been characterized in such networks. Many applications will require synchronized local clocks with varying levels of precision in order to maintain consistency and coordination in the network. Traditional time synchronization protocols however do not fit LoRaWAN's delay-inherent, low duty cycle, network model and wide-area deployment topology. Meanwhile, relying on GPS for time is not an option for low-power applications.

Learning requires both study and curiosity. A good learner is not only good at extracting information from the data given to it, but also skilled at finding the right new information to learn from. This is especially true when a human operator is required to provide the ground truth-such a source should only be queried sparingly. In this work, we address the problem of curiosity as it relates to online, real-time, human-in-the-loop training of an object detection algorithm onboard a robotic platform, one where motion produces new views of the subject. We propose a deep reinforcement learning approach that decides when to ask the human user for ground truth, and when to move. Through a series of experiments, we demonstrate that our agent learns a movement and request policy that is at least 3x more effective at using human user interactions to train an object detector than untrained approaches, and is generalizable to a variety of subjects and environments.

In this demonstration, we present OpenChirp, an open-source Low-Power Wide-Area Networking (LPWAN) infrastructure. OpenChirp is a management framework that provides data context, storage, visualization, and access control over the web. At the physical layer of the system, we present LPRAN, a low-cost high-performance software-defined radio hardware platform that can receive signals up to -30 dB below the noise floor. Using our LPRAN hardware, it is possible to operate on raw I/Q streams in the cloud to perform collaborative tasks across multiple gateways such as jointly decoding weak signals and localization.

Low-Power Wide-Area Networks (LPWANs) are an emerging wireless platform which can support battery-powered devices lasting 10-years while communicating at low data-rates to gateways several kilometers away. Not all such devices will experience the promised 10 year battery life despite the high density of LPWAN gateways expected in cities. Transmission from devices located deep within buildings or in remote neighborhoods will suffer severe attenuation forcing the use of slow data-rates to reach even the closest gateway, thus resulting in battery drain. This paper presents Charm, a system that enhances both the battery life of client devices and the coverage of LPWANs in large urban deployments. Charm allows multiple LoRaWAN gateways to pool their received signals in the cloud, coherently combining them to detect weak signals that are not decodable at any individual gateway. Through a novel hardware and software design at the gateway, Charm carefully detects which chunks of the received signal need to be sent to the cloud, thereby saving uplink bandwidth. We present a scalable solution to decoding weak transmissions at city-scale by identifying the set of gateways whose signals need to be coherently combined over time. In evaluations over a test network and from simulations using traces from a large LoRaWAN deployment in Pittsburgh, Pennsylvania, Charm demonstrates a gain of up to 3x in range and 4x in client battery-life.

Connected embedded systems in the realm of smart infrastructures comprise ubiquitous end-point devices supported by a communication infrastructure. Device, energy supply and network failures are a reality and provisioned communications could fail. Self-organization is a process where network devices cooperate with each other to restore network connectivity on detecting network connectivity failures. Self-organized networks are envisioned to be hierarchical, implying that a root device is expected to spend more energy to forward the entire network's data. This leads to battery exhaustion and therefore a single point of failure in the system. In this paper we address this problem by proposing an energy-governed resilient networking framework. Our framework enforces a policy to throttle upstream network traffic to maintain energy drain at the root device. To demonstrate the effectiveness of the proposed policy, we designed our experiment framework using Nano-RK and FireFly; a lightweight operating system and sensing platform respectively.