Detecting earthquakes in seismological time series is a core task in observational seismology, supporting a range of applications from early warning systems to tectonic research. Typically, seismic sensors passively record data and send it to the cloud or edge for integration, storage, and analysis. While this cloud-based approach is effective in urban or well-connected areas, it is impractical in remote, underwater, or underground environments where network infrastructure is unreliable. In such settings, the sensors must operate independently for extended periods while coping with strict constraints on power, memory, and connectivity. To address these challenges, we present LightEQ, a system that combines an efficient data processing pipeline and a lightweight deep-learning model specifically designed for seismic event detection in such environments. LightEQ runs on ultra-low-power microcontrollers with just 100 kB of RAM, enabling real-time, on-device earthquake detection without the need for continuous streaming of raw data to a central location. We evaluate LightEQ against a traditional STA/LTA approach and state-of-the-art (SOTA) machine learning models, using the Stanford Earthquake Dataset. Unlike existing neural network (NN) models, which are too large for microcontrollers, LightEQ is over ten times smaller than most of the SOTA models. Our results demonstrate that communication is the most energy-intensive task in this setting, and that traditional model-driven filters like STA/LTA are inefficient due to their high false positive rate. In contrast, LightEQ improves detection accuracy with NN, providing a more energy-efficient solution by reducing the number of false positives before transmission. Compared to the STA/LTA method alone, LightEQ extends battery life by at least 3-fold by minimizing energy consumption associated with transmitting false positives to the cloud.

In recent years, the use of machine learning (ML) in cellular networking has increased significantly, enabled by the availability of new ML algorithms and cellular datasets. However, existing 5G datasets focus mainly on land-based vehicular networks, which do not capture the unique challenges of the coastal maritime domain. This includes large distances from base stations, dynamic sea states, such as waves and tides, and varying interference from water surface reflections and nearby vessels. This paper introduces the Fjord5G11https://github.com/ds-kiel/Fjord5G, a 5G dataset for coastal maritime connectivity research. We conduct an extensive measurement campaign aboard research and public ferries in the Kiel Fjord, Germany, collecting GPS-located cellular data along maritime routes. These measurements cover the network conditions encountered in coastal and near-shore regions and provide insights into metrics such as signal strength, modulation, and bandwidth. The resulting dataset includes cellular measurements at a sampling rate of 1 Hz from two mobile network operators, four 5G routers, and two ferries for up to 12 months per router. Initial data analysis reveals key challenges for ML, such as dealing with varying bandwidth and handover events, while highlighting potential features, such as signal strength metrics, that can be exploited to improve coastal maritime connectivity.

Deploying large Deep Neural Networks with state-of-the-art accuracy on edge devices is often impractical due to their limited resources. This paper introduces EdgeBoost , a selective input offloading system designed to overcome the challenges of limited computational resources on edge devices. EdgeBoost trains and calibrates a lightweight model for deployment on the edge and, in addition, deploys a large, complex model on the cloud. During inference, the edge model makes initial predictions for input samples, and if the confidence of the prediction is low, the sample is sent to the cloud model for further processing, otherwise, we accept the local prediction. Through careful calibration, EdgeBoost reduces the communication cost by 55%, 27% and 20% for the CIFAR-100, ImageNet-1k and Stanford Cars datasets, respectively, when compared to an cloud-only solution while achieving on-par classification accuracy. Furthermore, EdgeBoost reduces the total inference latency from 148 ms to 123.84 ms per inference compared to a cloud-only solution. Our evaluation also shows that calibrating the edge model for such a collaborative edge-cloud setup results in accuracy gains of up to 8 percent point, compared to an uncalibrated edge model. Additionally, EdgeBoost, when used as an abstaining classifier, can improve accuracy by up to 9 percent points over an uncalibrated model. Finally, EdgeBoost outperforms the Early Exit and Entropy thresholding baselines and achieves comparable accuracy to state-of-the-art routing-based methods without the need for hosting the router on the edge.

In the context of cellular networks, such as with 5G and upcoming 6G networks, the available bandwidth of a connection is inherently dynamic. Accurate prediction of future bandwidth availability within a link is essential for latency-sensitive and mission-critical applications such as video streaming or remote driving. Bandwidth prediction ensures efficient utilization of a link and thus prevents delays. This paper introduces BandSeer, a stacked Bi-LSTM-based approach for bandwidth prediction in LTE and 5G cellular networks. BandSeer captures complex correlations in historical metrics better than prior work and outperforms SotA baselines. It achieves reductions of up to 18.32% in RMSE and 26.87% in MAE on the Berlin V2X dataset, and reductions of up to 12.43% in RMSE and 28.45% in MAE on the Beyond 5G dataset compared to the SotA Informer baseline. Furthermore, we argue that any bandwidth algorithm must be resource efficient to enable for development on various devices. Our evaluations show that BandSeer consumes one order of magnitude fewer resources and needs roughly a quarter to half the inference time of its closest competitor, the Informer model.

IoT devices have limited hardware capabilities and are often deployed in remote areas. Consequently, advanced vision models surpass such devices' processing and storage capabilities, requiring offloading of such tasks to the cloud. However, remote areas often rely on LPWANs technology with limited bandwidth, high packet loss rates, and extremely low duty cycles, which makes fast offloading for time-sensitive inference challenging. Today's approaches, which are deployable on weak devices, generate a non-progressive bit stream, and therefore, their decoding quality suffers strongly when data is only partially available on the cloud at a deadline due to limited bandwidth or packet losses.

The architecture of Vision Transformers (ViTs), particularly the Multi-head Attention (MHA) mechanism, imposes substantial hardware demands. Deploying ViTs on devices with varying constraints, such as mobile phones, requires multiple models of different sizes. However, this approach has limitations, such as training and storing each required model separately. This paper introduces HydraViT, a novel approach that addresses these limitations by stacking attention heads to achieve a scalable ViT. By repeatedly changing the size of the embedded dimensions throughout each layer and their corresponding number of attention heads in MHA during training, HydraViT induces multiple subnetworks. Thereby, HydraViT achieves adaptability across a wide spectrum of hardware environments while maintaining performance. Our experimental results demonstrate the efficacy of HydraViT in achieving a scalable ViT with up to 10 subnetworks, covering a wide range of resource constraints. HydraViT achieves up to 5 p.p. more accuracy with the same GMACs and up to 7 p.p. more accuracy with the same throughput on ImageNet-1K compared to the baselines, making it an effective solution for scenarios where hardware availability is diverse or varies over time. The source code is available at https://github.com/ds-kiel/HydraViT.

IoT devices rely on environment detection to trigger specific actions, e.g., for headphones to adapt noise cancellation to the surroundings. While phones feature many sensors, from GNSS to cameras, small wearables must rely on the few energy-efficient components they already incorporate. In this paper, we demonstrate that a Bluetooth radio is the only component required to accurately classify environments and present BlueSeer, an environment-detection system that solely relies on received BLE packets and an embedded neural network. BlueSeer achieves an accuracy of up to 84% differentiating between 7 environments on resource-constrained devices, and requires only ~ 12 ms for inference on a 64 MHz microcontroller-unit.

Testbeds have become a vital tool for evaluating and benchmarking applications and algorithms in the Internet of Things (IoT). Testbeds commonly consist of low-power IoT de-vices augmented with observer nodes providing control, logging, and often also power-profiling. Today, the research community operates numerous testbeds, sometimes with hundreds of IoT nodes, to allow for detailed and large-scale evaluation. Most testbeds, however, lack opportunities for tracing distributed program execution with high accuracy in time, for example, via minimally invasive, distributed GPIO tracing. And the ones that do, like Flocklab, are built from custom hardware, which is often too complex, inflexible, or expensive to use for other research groups.This paper closes this gap and introduces Grace, a low-cost, retrofittable, distributed, and time-synchronized GPIO tracing system built from off-the-shelf components, costing less than €20 per node. Grace extends observer nodes in a testbed with (1) time-synchronization via wireless sub-GHz transceivers and (2) logic analyzers for GPIO tracing and logging, enabling time-synchronized GPIO tracing at a frequency of up to 8 MHz. We deploy Grace in a testbed and show that it achieves an average time synchronization error between nodes of 1.53 μs.

With more than 4 billion devices produced in 2020, Bluetooth and Bluetooth Low Energy (BLE) have become the dominant solutions for short-range wireless communication in IoT. BLE mitigates interference via Adaptive Frequency Hopping (AFH), spreading communication over the entire spectrum. However, the ever-growing number of BLE devices and WiFi traffic in the already crowded 2.4 GHz band lead to situations where the quality of BLE connections dynamically changes with nearby wireless traffic, location, and time of day. These dynamic environments demand new approaches for channel management in AFH, by both dynamically excluding frequencies suffering from localized interference and adaptively re-including channels, thus providing sufficient channel diversity to survive the rise of new interference.We introduce eAFH, a new channel-management approach in BLE with a strong focus on efficient channel re-inclusion. eAFH introduces informed exploration as a driver for inclusion: using only past measurements, eAFH assesses which frequencies we are most likely to benefit from re-inclusion into the hopping sequence. As a result, eAFH adapts in dynamic scenarios where interference varies over time. We show that eAFH achieves 98-99.5% link-layer reliability in the presence of dynamic WiFi interference with 1% control overhead and 40% higher channel diversity than state-of-the-art approaches.

Deep Neural Networks (DNNs) on IoT devices are becoming readily available for classification tasks using sensor data like images and audio. However, DNNs are trained using extensive computational resources such as GPUs on cloud services, and once being quantized and deployed on the IoT device remain unchanged. We argue in this paper, that this approach leads to three disadvantages. First, IoT devices are deployed in real-world scenarios where the initial problem may shift over time (e.g., to new or similar classes), but without re-training, DNNs cannot adapt to such changes. Second, IoT devices need to use energy-preserving communication with limited reliability and network bandwidth, which can delay or restrict the transmission of essential training sensor data to the cloud. Third, collecting and storing training sensor data in the cloud poses privacy concerns. A promising technique to mitigate these concerns is to utilize on-device Transfer Learning (TL). However, bringing TL to resource-constrained devices faces challenges and trade-offs in computational, energy, and memory constraints, which this paper addresses. This paper introduces MicroTL, Transfer Learning (TL) on low-power IoT devices. MicroTL tailors TL to IoT devices without the communication requirement with the cloud. Notably, we found that the MicroTL takes 3x less energy and 2.8x less time than transmitting all data to train an entirely new model in the cloud, showing that it is more efficient to retrain parts of an existing neural network on the IoT device.

Bluetooth Low Energy (BLE) is the prevalent IoT radio technology and perfectly suited for mobile and battery-driven applications. However, it is not designed for intermittent connectivity and opportunistic networking. Hence, the vast infrastructure that BLE-equipped devices such as smartphones, wearables, and sensors provide, remains untapped, even with its potential for data collection, sharing, or emergency communication in disaster scenarios. This paper introduces DISRUPTABLE to unfold this potential: A universal BLE-based store-and-forward architecture for delay-tolerant and opportunistic networking. Tailored to the resource constraints of IoT nodes and the feature set of BLE, DISRUPTABLE enables opportunistic interactions between BLE-equipped devices, providing a resilient network even when established communication over cellular networks or Wi-Fi fails. In our evaluation, we show that in a highly dynamic pedestrian scenario in downtown Stockholm, broadcasts reliably inform pedestrians in 7.1 seconds, while unicast messages arrive within 20 minutes in 48.1% of cases.

The last decade saw an emergence of Synchronous Transmissions (ST) as an effective communication paradigm in low-power wireless networks. Numerous ST protocols provide high reliability and energy efficiency in normal wireless conditions, for a large variety of traffic requirements. Recently, with the EWSN dependability competitions, the community pushed ST to harsher and highly-interfered environments, improving upon classical ST protocols through the use of custom rules, hand-tailored parameters, and additional retransmissions. The results are sophisticated protocols, that require prior expert knowledge and extensive testing, often tuned for a specific deployment and envisioned scenario. In this paper, we explore how ST protocols can benefit from self-adaptivity; a self-adaptive ST protocol selects itself its best parameters to (1) tackle external environment dynamics and (2) adapt to its topology over time. We introduce Dimmer as a self-adaptive ST protocol. Dimmer builds on LWB and uses Reinforcement Learning to tune its parameters and match the current properties of the wireless medium. By learning how to behave from an unlabeled dataset, Dimmer adapts to different interference types and patterns, and is able to tackle previously unseen interference. With Dimmer, we explore how to efficiently design AI-based systems for constrained devices, and outline the benefits and downfalls of AI-based low-power networking. We evaluate our protocol on two deployments of resource-constrained nodes achieving 95.8 % reliability against strong, unknown WiFi interference. Our results outperform baselines such as non-adaptive ST protocols (~27%) and PID controllers, and show a performance close to hand-crafted and more sophisticated solutions, such as Crystal (~99 %).

Low-power wireless networking commonly uses either Time-Slotted Channel Hopping (TSCH), synchronous transmissions, or opportunistic routing. All three of these different, orthogonal approaches strive for efficient and reliable communication but follow different trajectories. With this paper, we combine these concepts into one protocol: Autobahn.Autobahn merges TSCH scheduling with opportunistically routed, synchronous transmissions. This opens the possibility to create long-term stable schedules overcoming local interference. We prove the stability of schedules over several days in our experimental evaluation. Moreover, Autobahn outperforms the autonomous scheduler Orchestra under interference in terms of reliability by 13.9 percentage points and in terms of latency by a factor of 9 under a minor duty cycle increase of 2.1 percentage points.

Bluetooth is an omnipresent technology, available on billions of devices today. While it has been traditionally limited to peer-to-peer communication and star networks, the recent Bluetooth Mesh standard extends it to multi-hop networking. In addition, the Bluetooth 5 standard introduces new modes to allow for increased reliability. In this article, we evaluate the feasibility of concurrent transmissions (CT) in Bluetooth via modeling and controlled experiments and then devise an efficient network-wide data dissemination protocol, BlueFlood, based on CT for multi-hop Bluetooth networks. First, we model and analyze how CT distorts the received waveform and characterize the Bit Error Rate of a Frequency-Shift Keying receiver to show that CT is feasible over Bluetooth. Second, we verify our analytic results with a controlled experimental study of CT over Bluetooth PHY. Third, we present BlueFlood, a fast and efficient network-wide data dissemination in multi-hop Bluetooth networks. In our experimental evaluation, in two testbeds deployed in university buildings, we show that BlueFlood achieves 99.9% end-to-end delivery ratio with a duty-cycle of 0.4% for periodic dissemination of advertising packets of 38 bytes with 200 milliseconds intervals at 2 Mbps. Moreover, we show that BlueFlood can be received by off-the-shelf devices such as smartphones, paving a seamless integration with existing technologies.

Wireless Sensor-Actuator Networks (WSANs) are an important driver for the Industrial Internet of Things (IIoT) as they easily retrofit existing industrial infrastructure. Industrial applications require these networks to provide stable communication with high reliability and guaranteed low latency. A common way is using a central scheduler to plan transmissions and routes so that all packets are delivered before a deadline. However, existing centralized schedulers are only able to achieve high reliability in the absence of interference. This limitation lowers the feasibility of using centralized schedulers in most environments susceptible to interference.This paper addresses the challenge of stable, centrally scheduled communication in low-power wireless networks susceptible to interference. We introduce MASTER, a centralized scheduler and router, for IEEE 802.15.4 TSCH (Time-Slotted Channel Hopping). MASTER uses Sliding Windows, a novel transmission strategy, which builds on flow-based retransmissions instead of link-based ones. We show in our experimental evaluation that MASTER with Sliding Windows achieves routing and scheduling stability for over 24 hours with end-to-end reliability of over 99.6%. Moreover, we show that MASTER outperforms Orchestra, a state-of-the-art autonomous scheduler, in terms of latency by a factor of 8 while achieving similar reliability under a slight duty-cycle increase.

In this paper, we present Overtake, an opportunistic routing protocol for Time-Slotted Channel Hopping (TSCH). Overtake combines (1) opportunistic routing, (2) concurrent transmissions and (3) TSCH. We show that this novel combination enables low-latency, central scheduling withstanding node failures. Our initial results show its ability to withstand node failures of up to 40% of nodes of a flow while keeping minimal latency.

Sketches are data structures designed to answer approximate queries by trading memory overhead with accuracy guarantees. More specifically, sketches efficiently summarize large, high-rate streams of data and quickly answer queries on these summaries. In order to support such high throughput rates in modern architectures, parallelization and support for fast queries play a central role, especially when monitoring unpredictable data that can change rapidly as, e.g., in network monitoring for large-scale denial-of-service attacks. However, most existing parallel sketch designs have focused either on high insertion rate or on high query rate, and fail to support cases when these operations are concurrent.

With the rise of the Internet of Things (IoT), billions of devices ranging from simple sensors to smart-phones will participate in billions of micropayments. However, current centralized solutions are unable to handle a massive number of micropayments from untrusted devices. Blockchains are promising technologies suitable for solving some of these challenges. Particularly, permissionless blockchains such as Ethereum and Bitcoin have drawn the attention of the research community. However, the increasingly large-scale deployments of blockchain reveal some of their scalability limitations. Prominent proposals to scale the payment system include off-chain protocols such as payment channels. However, the leading proposals assume powerful nodes with an always-on connection and frequent synchronization. These assumptions require in practice significant communication, memory, and computation capacity, whereas IoT devices face substantial constraints in these areas. Existing approaches also do not capture the logic and process of IoT, where applications need to process locally collected sensor data to allow for full use of IoT micro-payments. In this paper, we present TinyEVM, a novel system to generate and execute off-chain smart contracts based on sensor data. TinyEVM's goal is to enable IoT devices to perform micro-payments and, at the same time, address the device constraints. We investigate the trade-offs of executing smart contracts on low-power IoT devices using TinyEVM. We test our system with 7,000 publicly verified smart contracts, where TinyEVM achieves to deploy 93% of them without any modification. Finally, we evaluate the execution of off-chain smart contracts in terms of run-time performance, energy, and memory requirements on IoT devices. Notably, we find that low-power devices can deploy a smart contract in 215 ms on average, and they can complete an off-chain payment in 584 ms on average.

Intra-group spreading is one of the common and frequent needs in many decentralized communication protocols. Such spreading is useful for quick and local dissemination of information among different clusters in large-scale decentralized systems like the Internet-of-Things (IoT). Complex decentralized protocols can carefully exploit such localized dissemination as a base unit for their efficient implementation. However, due to the inherent broadcast nature of wireless communication, efficient and simultaneous execution of multiple intra-group disseminations is difficult. In this work, we propose a novel and simple way to completely hide a wireless transmission without changing any channel or frequency. Next, we use this for supporting simultaneous intra-group disseminations. Rigorous evaluation of the proposed strategy over testbeds show significant improvement of upto 60% in reliability with similar average latency and radio-on time in comparison to the baseline where no additional mechanism is adopted for separation of intra-group communications.

Intersections are the bottlenecks of road networks. Coordination mechanisms for intersection crossing greatly affect the efficiency of road utilization. Typically, coordination is done by implanting local infrastructure, whether signs, traffic lights, or through common, well known-rules shared by all users. In this paper, we introduce STARC, a decentralized intersection management protocol for future connected vehicles and other traffic participants. With STARC, all participants coordinate their movement using reservations to guarantee safe crossings. To enable cost-efficient deployment, STARC does neither rely on any centralized infrastructure, such as traffic lights, nor centralized wireless intersection coordinators, like virtual traffic lights. STARC targets small, cheap, and energy-efficient platforms and the open low-power wireless standard 802.15.4 so that all participants in road traffic could take advantage of it, including vehicles, bikes, electric scooters, and even pedestrians. STARC builds on low-power wireless communication with A2-Synchrotron and multi-hop routing as a communication substrate and provides distributed transaction, election, and handover mechanisms to manage the intersection cooperatively. We show that STARC reduces average waiting times by up to 50% compared to a fixed traffic light schedule in traffic volumes with less than 1000 vehicles per hour. Moreover, we illustrate a platoon extension that allows STARC to outperform traffic lights even at traffic loads beyond 1000 vehicles per hour.

Pattern matching is an important building block for many security applications, including Network Intrusion Detection Systems (NIDS). As NIDS grow in functionality and complexity, the time overhead and energy consumption of pattern matching become a significant consideration that limits the deployability of such systems, especially on resource-constrained devices. On the other hand, the emergence of new computing platforms, such as embedded devices with integrated, general-purpose Graphics Processing Units (GPUs), brings new, interesting challenges and opportunities for algorithm design in this setting: how to make use of new architectural features and how to evaluate their effect on algorithm performance. Up to now, work that focuses on pattern matching for such platforms has been limited to specific algorithms in isolation.

With the proliferation of the Internet of Things (IoT), the need to prioritize the overall system security is more imperative than ever. The IoT will profoundly change the established usage patterns of embedded systems, where devices traditionally operate in relative isolation. Internet connectivity brought by the IoT exposes such previously isolated internal device structures to cyber-attacks through the Internet, which opens new attack vectors and vulnerabilities. For example, a malicious user can modify the firmware or operating system by using a remote connection, aiming to deactivate standard defenses against malware. The criticality of applications, for example, in the Industrial IoT (IIoT) further underlines the need to ensure the integrity of the embedded software. One common approach to ensure system integrity is to verify the operating system and application software during the boot process. However, safety-critical IoT devices have constrained boot-up times, and home IoT devices should become available quickly after being turned on. Therefore, the boot-time can affect the usability of a device. This paper analyses performance trade-offs of secure boot for medium-scale embedded systems, such as Beaglebone and Raspberry Pi. We evaluate two secure boot techniques, one is only software-based, and the second is supported by a hardware-based cryptographic storage unit. For the software-based method, we show that secure boot merely increases the overall boot time by 4%. Moreover, the additional cryptographic hardware storage increases the boot-up time by 36%.

Continuously monitoring sensor readings is an important building block for many IoT applications. The literature offers resourceful methods that minimize the amount of communication required for continuous monitoring, where Geometric Monitoring (GM) is one of the most generally applicable ones. However, GM has unique communication requirements that require specialized network protocols to unlock the full potential of the algorithm. In this work, we show how application and protocol co-design can improve the real-life performance of GM, making it an application of practical value for real IoT deployments. We orchestrate the communication of GM to utilize the properties of a state-of-the-art wireless protocol (Crystal) that relies on synchronous transmissions and is designed for aperiodic traffic, as needed by GM. We bridge the existing gap between the capabilities of the protocol and the requirements of GM, especially in the case of periods of heavy communication. We do so by introducing an in-network aggregation technique relying on latent opportunities for aggregation that we exploit in Crystal's design, allowing us to reliably monitor duplicate-sensitive aggregate functions, such as sum, average or variance. Our results from testbed experiments with a publicly available dataset show that the combination of GM and Crystal results in a very small duty-cycle, a 2.2x - 3.2x improvement compared to the baseline and up to 10x compared to previous work. We also show that our in-network aggregation technique reduces the duty-cycle by up to 1.38x.

For any distributed system, and especially for the Internet of Things, recording interactions between devices is essential. At first glance, blockchains seem to be suitable for storing these interactions, as they allow multiple parties to share a distributed ledger. However, at a closer look, blockchains require heavy computations, large memory capacity, and always-on communication to the cloud; these are three properties that are challenging for IoT devices with limited resources. In this paper, we present IoTLogBlock to address these challenges. IoTLogBlock connects resource-constrained IoT devices to the blockchain, and it consists of three building blocks jointly enabling recording transactions: a lightweight contract signing protocol, a blockchain network, and a smart contract. The contract signing protocol allows devices to interact locally to perform transactions, even if no communication to the cloud and the blockchain exists at that moment. At a later time, devices forward the stored transactions to the blockchain, where a smart contract ultimately verifies the transactions. We evaluate our design on low-power devices and quantify the performance in terms of memory, computation, and energy consumption. Our results show that a constrained device can create and sign a transaction within 3 s on average. Finally, we expose the devices to network scenarios with edge connections ranging from 10 s to over 2 h.

This article presents Whisper, a fast and reliable protocol to flood small amounts of data into a multi-hop network. Whisper makes use of synchronous transmissions, a technique first introduced by the Glossy flooding protocol. In contrast to Glossy, Whisper does not let the radio switch from receive to transmit mode between messages. Instead, it makes nodes continuously transmit identical copies of the message and eliminates the gaps between subsequent transmissions. To this end, Whisper embeds the message to be flooded into a signaling packet that is composed of multiple packlets —where a packlet is a portion of the message payload that mimics the structure of an actual packet. A node must intercept only one of the packlets to detect that there is an ongoing transmission and that it should start forwarding the message. This allows Whisper to speed up the propagation of the flood and, thus, to reduce the overall radio-on time of the nodes. Our evaluation on the FlockLab testbed shows that Whisper achieves comparable reliability but 2× lower radio-on time than Glossy. We further show that by embedding Whisper in an existing data collection application, we can more than double the lifetime of the network.

IoT deployments often require communication between devices that employ heterogeneous wireless technologies. Traditionally, expensive gateways are used to relay packets between heterogeneous nodes. Recent cross-technology communication offers a low bandwidth alternative, which is only feasible when communication between such nodes is limited to simple binary commands. In contrast, our work capitalizes on the increasing presence of multi-standard radio chips in mainstream IoT devices, to provide a new perspective on how to enable direct communication between heterogeneous nodes. We design Scylla---a software control layer---that allows multiple wireless stacks to coexist on top of a single radio chip, thereby simultaneously offering multiple communication interfaces. Uniquely, Scylla achieves near stack-native performance and requires no changes to the standards.

Applications for IoT often continuously monitor sensor values and react if the network-wide aggregate exceeds a threshold. Previous work on Geometric monitoring (GM) has promised a several-fold reduction in communication but been limited to analytic or high-level simulation results. In this paper, we build and evaluate a full system design for GM on resource-constrained devices. In particular, we provide an algorithmic implementation for commodity IoT hardware and a detailed study regarding duty cycle reduction and energy savings. Our results, both from full-system simulations and a publicly available testbed, show that GM indeed provides several-fold energy savings in communication. We see up to 3x and 11x reduction in duty-cycle when monitoring the variance and average temperature of a real-world data set, but the results fall short compared to the reduction in communication (4.3x and 44x, respectively). Hence, we investigate the energy overhead imposed by the network stack and the communication pattern of the algorithm and summarize our findings. These insights may enable the design of protocols that will unlock more of the potential of GM and similar algorithms for IoT deployments.

Achieving fast and efficient many-to-many communication is one of the most complex communication problems, especially in wireless systems. A compact form of many-to-many communication in a distributed system has the potential to bring huge benefit to many distributed algorithms and protocols. Many-to-many communication can be implemented as a sequential instantiations of a network wide one-to-many communication. One limitation of such an approach is that each individual instance of a one-to-many communication has to be given enough time to propagate through the whole network before the next instance. In addition, there is large overhead in generating the schedule for the sequence of individual one-to-many communications. In this work, we show that many-to-many communication can be more efficiently implemented as many parallel many-to-one communications. In this direction, we first develop an efficient TDMA based many-to-one communication module, and then use it in many-to-many setting. Our approach achieves a minimum about 20% to 50% improvements on latency (radio-on time) over the state-of-the-art solutions in a 90-node wireless sensor network testbed.

Pattern matching is a key building block of Intrusion Detection Systems and firewalls, which are deployed nowadays on commodity systems from laptops to massive web servers in the cloud. In fact, pattern matching is one of their most computationally intensive parts and a bottleneck to their performance. In Network Intrusion Detection, for example, pattern matching algorithms handle thousands of patterns and contribute to more than 70% of the total running time of the system.In this paper, we introduce efficient algorithmic designs for multiple pattern matching which (a) ensure cache locality and (b) utilize modern SIMD instructions. We first identify properties of pattern matching that make it fit for vectorization and show how to use them in the algorithmic design. Second, we build on an earlier, cache-aware algorithmic design and we show how cache-locality combined with SIMD gather instructions, introduced in 2013 to Intel's family of processors, can be applied to pattern matching. We evaluate our algorithmic design with open data sets of real-world network traffic:Our results on two different platforms, Haswell and Xeon-Phi, show a speedup of 1.8x and 3.6x, respectively, over Direct Filter Classification (DFC), a recently proposed algorithm by Choi et al. for pattern matching exploiting cache locality, and a speedup of more than 2.3x over Aho-Corasick, a widely used algorithm in today's Intrusion Detection Systems.

Emerging protocols for low-power wireless networks increasingly exploit constructive interference and the capture effect. The basic idea is that the synchronous transmission of identical packets by neighboring nodes leads to constructive interference - or at least do not cause destructive interference. This requires that the temporal displacement of packets at receiving nodes is lower than 0.5 μs when employing IEEE 802.15.4 radios. However, commonly used sensor nodes are equipped with cheap and imprecise clocks that show high frequency deviations across nodes, making constructive interference difficult to achieve. Such deviations further increase when individual nodes are exposed to different temperatures. In this paper we introduce Flock, a novel approach to compensate for differences in clock frequency across synchronously transmitting nodes. We implemented Flock in Contiki on the example of Glossy, a flooding protocol based on synchronous transmissions. Our results confirm that Flock can achieve constructive interference on real sensor nodes in over 98% of the cases. Overall, Flock makes protocols that exploit synchronous transmissions more robust to operate even in challenging environments.

We propose dynamic reconfiguration of the radio interface in IoT platforms to support multiple link layers simultaneously. This allows us to tackle the increasing link layer heterogeneity in IoT devices, thus bringing a multitude of benefits: extensible and vender agnostic multihop deployments, rapid integration of the new “things” into an existing network, as well as seamless integration of the IoT with the traditional wireless Internet.

In low-power wireless networking, new applications such as cooperative robots or industrial closed-loop control demand for network-wide consensus at low-latency and high reliability. Distributed consensus protocols is a mature field of research in a wired context, but has received little attention in low-power wireless settings. In this paper, we present A2: Agreement in the Air, a system that brings distributed consensus to low-power multi-hop networks. A2 introduces Synchrotron, a synchronous transmissions kernel that builds a robust mesh by exploiting the capture effect, frequency hopping with parallel channels, and link-layer security. A2 builds on top of this reliable base layer and enables the two- and three-phase commit protocols, as well as network services such as group membership, hopping sequence distribution and re-keying.

Many protocols in low-power wireless networks require a root node or a leader to bootstrap and maintain its operation. For example, Chaos and Glossy networks need an initiator to synchronize and initiate the communications rounds. Commonly, these protocols use a fixed, compile-time defined node as the leader. In this work, we tackle the challenge of dynamically bootstrapping the network and electing a leader in low-power wireless scenarios, and we focus on Chaos-style networks.

The introduction of Advanced Metering Infrastructures in electricity networks brings new means of dealing with issues influencing financial margins and system-safety problems, thanks to the information reported continuously by smart meters. Such an issue is the detection of Non-Technical Losses (NTLs) in electric power grids. We introduce a data-driven method, called Structure&Detect, to identify possible sources of NTLs; the method is based on spectral analysis of structural periodic patterns in consumption traces, that allows for scalable processing, using features in the frequency domain. Structure&Detect uses only on consumption traces, with no need for exogenous data about customers (e.g., trust or credit history) or explicit information from domain experts. As such, it complies better with privacy concerns that may be present when processing data from different sources. Using real-world consumption traces, we show that it provides high accuracy and detection rates comparable to methods that require additional, customer-specific information. Moreover, Structure&Detect can also be used orthogonally due to its high detection rate, as a filter, providing a narrowed-down input set to methods requiring different treatment (e.g. additional data or on-site inspection) and thus make the search for NTLs more scalable. Structure&Detect also enables processing each meter trace on-the-fly, as well as in a parallel and distributed fashion. These properties make Structure&Detect suitable for online analysis that can address common big data challenges such as the need for scalable, distributed and parallel analysis close to IoT edge devices, such as smart meters.

Standard Internet communication protocols are key enablers for the Internet of Things (IoT). Recent technological advances have made it possible to run such protocols on resource-constrained devices. Yet these devices often use energy-efficient, low-level communication technologies, like IEEE 802.15.4, which suffer from low-reliability and high latency. These drawbacks can be significantly reduced if communication occurs using concurrent transmissions - a novel communication paradigm for resource-constrained devices. In this paper, we show that Internet protocols like TCP/UDP and CoAP can run efficiently on top of a routing substrate based on concurrent transmissions. We call this substrate LaneFlood and demonstrate its effectiveness through extensive experiments on Flocklab, a publicly available testbed. Our results show that LaneFlood improves upon CXFS - a representative competitor - in terms of both duty cycle and reliability. Furthermore, LaneFlood can transport IoT traffic with an end-to-end latency of less than 300 ms over several hops.

Experimental research in low-power wireless networking lacks a reference benchmark. While other communities such as databases or machine learning have standardized benchmarks, our community still uses ad-hoc setups for its experiments and struggles to provide a fair comparison between communication protocols. Reasons for this include the diversity of network scenarios and the stochastic nature of wireless experiments. Leveraging on the excellent testbeds and tools that have been built to support experimental validation, we make the case for a reference benchmark to promote a fair comparison and reproducibility of results. This abstract describes early design elements and a benchmarking methodology with the goal to gather feedback from the community rather than propose a definite solution.

Time slotted operation is a well-proven approach to achieve highly reliable low-power networking through scheduling and channel hopping. It is, however, difficult to apply time slotting to dynamic networks as envisioned in the Internet of Things. Commonly, these applications do not have pre-defined periodic traffic patterns and nodes can be added or removed dynamically.

Cyber-Physical Systems (CPS) employ sensors to observe physical environments and to detect events of interest. Equipped with sensing, computing, and communication capabilities, Cyber-Physical Systems aim to make physical-systems smart(er). For example, smart electricity meters nowadays measure and report power consumption as well as critical events such as power outages. However, each day, such sensors report a variety of warnings and errors: many merely indicate transient faults or short instabilities of the physical system (environment). Thus, given the big volumes of data, the time-efficient processing of these events, especially in large-scale scenarios with hundreds of thousands of sensors, is a key challenge in CPSs. Motivated by the fact that critical events of CPSs often have temporal-spatial properties, we focus on identifying critical events by an online temporal-spatial analysis on the data stream of messages. We explicitly model the online detection problem as a single-linkage clustering on a data stream over a sliding-window, where the inherent computational complexity of the detection problem is derived. Based on this model, we propose a grid-based single-linkage clustering algorithm over a sliding-window, which is an online time-space efficient method satisfying the quick processing demand of big data streams. We analyze the performance of the proposed approach by both a series of propositions and a large, real-world data-set of deployed CPS, composing 300,000 sensors, over one year. We show that the proposed method identifies above 95% of the critical events in the data-set and save the time-space requirement by 4 orders of magnitude compared with the conventional clustering method.

Low Power Listening (LPL) MAC protocols are widely used in today's sensors networks for duty cycling. Their simplicity and power efficiency ensures a long network life when nodes are battery driven and their easy deployment and lower cost of maintenance makes them suitable to be used in hard-to-access places and harsh conditions. We argue that to fully utilize energy efficiency provided by LPL, other protocols in the protocol stack should be aware of mechanisms. In this paper, we focus on neighborhood discovery protocols and discuss their energy efficient integration with LPL. Then, we study the possibility of using a completely passive approach for neighborhood discovery in such networks and provide an analytical model for its performance characteristics. We verify our performance model both by simulation and implementation in TinyOS. Our evaluation results confirm the efficiency of our proposed method in duty-cycled sensor networks.

Opportunistic routing is widely known to have substantially better performance than unicast routing in wireless networks with lossy links. However, wireless sensor networks are usually duty cycled, that is, they frequently enter sleep states to ensure long network lifetime. This renders existing opportunistic routing schemes impractical, as they assume that nodes are always awake and can overhear other transmissions. In this article we introduce ORW, a practical opportunistic routing scheme for wireless sensor networks. ORW uses a novel opportunistic routing metric, EDC, that reflects the expected number of duty-cycled wakeups that are required to successfully deliver a packet from source to destination. We devise distributed algorithms that find the EDC-optimal forwarding and demonstrate using analytical performance models and simulations that EDC-based opportunistic routing results in significantly reduced delay and improved energy efficiency compared to traditional unicast routing. In addition, we evaluate the performance of ORW in both simulations and testbed-based experiments. Our results show that ORW reduces radio duty cycles on average by 50% (up to 90% on individual nodes) and delays by 30% to 90% when compared to the state-of-the-art.

Routing in battery-operated wireless networks is challenging, posing a tradeoff between energy and latency. Previous work has shown that opportunistic routing can achieve low-latency data collection in duty-cycled networks. However, applications are now considered where nodes are not only periodic data sources, but rather addressable end points generating traffic with arbitrary patterns.

An important building block for low-power wireless systems is to efficiently share and process data among all devices in a network. However, current approaches typically split such all-to-all interactions into sequential collection, processing, and dissemination phases, thus handling them inefficiently.

Despite the continuous research to facilitate WSNs development, most safety analysis and mitigation efforts in concurrency are still left to developers, who must manage synchronization and shared memory explicitly. In this paper, we present a system language that ensures safe concurrency by handling threats at compile time, rather than at runtime. Based on the synchronous programming model, our design allows for a simple reasoning about concurrency that enables compile-time analysis resulting in deterministic and memory-safe programs. As a trade-off, our design imposes limitations on the language expressiveness, such as doing computationally-intensive operations and meeting hard real-time responsiveness. To show that the achieved expressiveness and responsiveness is sufficient for a wide range of WSN applications, we implement widespread network protocols and the CC2420 radio driver. The implementations show a reduction in source code size, with a penalty of memory increase below 10% in comparison to nesC. Overall, we ensure safety properties for programs relying on high-level control abstractions that also lead to concise and readable code.

Multichannel communication has been proposed as alternative to adaptive (single-channel) routing protocols for mitigating the impact of interference and link dynamics in wireless sensor networks. While several studies have advocated features of both techniques (not without running up against contradicting arguments) a comprehensive study that aligns these results is still lacking. This paper aims at filling this gap. We present an experimental test bed setup used to perform extensive measurements for both single-channel and multichannel communication. We first analyze single-channel and multichannel communication over a single-hop in terms of packet reception ratio, maximum burst loss, temporal correlation of losses, and loss correlations across channels. Results show that multichannel communication with channel hopping significantly reduces link burstiness and packet loss correlation. For multi-hop networks, multi-channel communication and adaptive routing show similar end-to-end reliability in dense topologies, while multichannel communication can outperform adaptive routing in sparse networks with bursty links.

The main task of most deployed wireless sensor networks is data collection. While a number of solutions have been designed for static networks, there are currently no widely used data collection algorithms for mobile sensor networks. In this paper, we concentrate on scenarios where many nodes, both data sources and sinks, move along a certain track in one direction, a scenario that is common in sports events. Rather than designing a new protocol from scratch, we extend an existing data collection protocol with lightweight mechanisms to make it efficient for mobility. Our extensive simulations and results in a test bed that includes mobile robots demonstrate that our solution is able to achieve high packet delivery rates at low energy consumption. For our target scenario, our solution more than doubles packet delivery rates when the network is sparse. Our solution also works well in scenarios with a higher degree of mobility where nodes move according to a more demanding random waypoint model.

Interference and link dynamics constitute great concerns for stability and performance of protocols in WSNs. In this paper we evaluate the impact of channel hopping and adaptive routing on delay and reliability focusing on delay critical applications.

Traditionally, routing in wireless sensor networks consists of two steps: First, the routing protocol selects a next hop, and, second, the MAC protocol waits for the intended destination to wake up and receive the data. This design makes it difficult to adapt to link dynamics and introduces delays while waiting for the next hop to wake up.

Opportunistic routing is widely known to have substantially better performance than traditional unicast routing in wireless networks with lossy links. However, wireless sensor networks are heavily duty-cycled, i.e. they frequently enter deep sleep states to ensure long network life-time. This renders existing opportunistic routing schemes impractical, as they assume that nodes are always awake and can overhear other transmissions. In this paper, we introduce a novel opportunistic routing metric that takes duty cycling into account. By analytical performance modeling and simulations, we show that our routing scheme results in significantly reduced delay and improved energy efficiency compared to traditional unicast routing. The method is based on a new metric, EDC, that reflects the expected number of duty cycled wakeups that are required to successfully deliver a packet from source to destination. We devise distributed algorithms that find the EDC-optimal forwarding, i.e. the optimal subset of neighbors that each node should permit to forward its packets. We compare the performance of the new routing with ETX-optimal single path routing in both simulations and testbed-based experiments.

Wireless control systems consist of sensing and actuating devices that are commonly driven by a central controller. Wireless communication protocols for Cyber-Physical Systems (CPS) match this design by employing a "sense → collect → process → disseminate → actuate" flow [6], where typically different protocols are employed for collecting sensor data and disseminating actuation signals.

Long-term deployments of sensor networks in physically inaccessible environments make remote re-programmability of sensor nodes a necessity. Ranging from full image replacement to virtual machines, a variety of mechanisms exist today to deploy new software or to fix bugs in deployed systems. However, TinyOS - the current state of the art sensor node operating system - is still limited to full image replacement as nodes execute a statically-linked system-image generated at compilation time. In this paper we introduce Dynamic TinyOS to enable the dynamic exchange of software components and thus incrementally update the operating system and its applications. The core idea is to preserve the modularity of TinyOS, i.e.~its componentization, which is lost during the normal compilation process, and enable runtime composition of TinyOS components on the sensor node. The proposed solution integrates seamlessly into the system architecture of TinyOS: It does not require any changes to the programming model of TinyOS and existing components can be reused transparently. Our evaluation shows that Dynamic TinyOS incurs a low performance overhead while keeping a smaller - upto one third - memory footprint than other comparable solutions.

We present Statistical Vector Routing (SVR), a protocol that efficiently deals with communication link dynamics in wireless networks. It assigns virtual coordinates to nodes based on the statistical distribution of their distance from a small set of beacons. The distance metric predicts the current location of a node in its address distribution. Our initial results from a prototype implementation over real testbeds demonstrate the feasibility of SVR.

Complex interactions and the distributed nature of wireless sensor networks make automated testing and debugging before deployment a necessity. A main challenge is to detect bugs that occur due to non-deterministic events, such as node reboots or packet duplicates. Often, these events have the potential to drive a sensor network and its applications into corner-case situations, exhibiting bugs that are hard to detect using existing testing and debugging techniques.

The simulation models of wireless networks rapidly increase in complexity to accurately model wireless channel characteristics and the properties of advanced transmission technologies. Such detailed models typically lead to a high computational load per simulation event that accumulates to extensive simulation runtimes. Reducing runtimes through parallelization is challenging since it depends on detecting causally independent events that can execute concurrently. Most existing approaches base this detection on lookaheads derived from channel propagation latency or protocol characteristics. In wireless networks, these lookaheads are typically short, causing the potential for parallelization and the achievable speedup to remain small. This paper presents Horizon, which unlocks a substantial portion of a simulation model's workload for parallelization by going beyond the traditional lookahead. We show how to augment discrete events with durations to identify a much larger horizon of independent simulation events and efficiently schedule them on multi-core systems. Our evaluation shows that this approach can significantly cut down the runtime of simulations, in particular for complex and accurate models of wireless networks.

Link estimation provides a long-term estimate of the quality of a link based on its past history. However, this need for a history of past packets is also its main drawback: First, most link estimators only adapt slowly to changing link conditions, being mainly designed to identify long-term stable links. As a result they leave out bursty, potentially long ranging links [2, 4]. Second, in low traffic environments, as seen in many of today's typically heavily duty-cycled application-slink estimates are potentially outdated as they are based on old packets. Finally, it requires to store estimates, i.e., state information, for its neighbors, and (4) relies on beacons to probe links.

This paper presents Horizon, an extension to discrete event simulation to accurately and efficiently model timing behavior. Horizon achieves high run-time performance by utilizing timing information for horizontal parallelization. We additionally illustrated the viability of Horizon by means of a prototype implementation and evaluation.

Accurate estimation of link quality is the key to enable efficient routing in wireless sensor networks. Current link estimators focus mainly on identifying long-term stable links for routing. They leave out a potentially large set of intermediate links offering significant routing progress. Fine-grained analysis of link qualities reveals that such intermediate links are bursty, i.e., stable in the short term.

The rising complexity of data processing algorithms in sensor networks combined with their severely limited computing power necessitates a in-depth understanding of their temporal behavior. However, today only cycle accurate emulation and test-beds provide a detailed and accurate insight into the temporal behavior of sensor networks.In this paper we introduce fine grained, automated instrumentation of simulation models with cycle counts derived from sensor nodes and application binaries to provide detailed timing information. The presented approach bridges the gap between scalable but abstracting simulation and cycle accurate emulation for sensor network evaluation.By mapping device-specific code with simulation models, we can derive the time and duration a certain code line takes to get executed on a sensor node. Hence, eliminating the need to use expensive instruction-level emulators with limited speed and restricted scalability. Furthermore, the proposed design is not bound to a specific hardware platform, a major advantage compared to existing emulators. Our evaluation shows that the proposed technique achieves a timing accuracy of 99% compared to emulation while adding only a small overhead. Concluding, it combines essential properties like accuracy, speed and scalability on a single simulation platform.

Although recent years provided many protocols for anonymous routing in overlay networks, they commonly rely on the same communication paradigm: Onion Routing. In Onion Routing a static tunnel through an overlay network is build via layered encryption. All traffic exchanged by its end points is relayed through this tunnel. In contrast, this paper introduces dynamic multipath Onion Routing to extend the static Onion Routing paradigm. This approach allows each packet exchanged between two end points to travel along a different path. To provide anonymity the first half of this path is selected by the sender and the second half by the receiver of the packet. The results are manifold: First, dynamic multipath Onion Routing increases the resilience against threats, especially pattern and timing based analysis attacks. Second, the dynamic paths reduce the impact of misbehaving and overloaded relays. Finally, inspired by Internet routing, the forwarding nodes do not need to maintain any state about ongoing flows and so reduce the complexity of the router. In this paper, we describe the design of our dynamic Multipath Onion RoutEr (MORE) for peer-to-peer overlay networks, and evaluate its performance. Furthermore, we integrate address virtualization to abstract from Internet addresses and provide transparent support for IP applications. Thus, no application-level gateways, proxies or modifications of applications are required to sanitize protocols from network level information. Acting as an IP-datagram service, our scheme provides a substrate for anonymous communication to a wide range of applications using TCP and UDP.

For the use in the Internet domain, distributed hash tables (DHTs) have proven to be an efficient and scalable approach to distributed content storage and access. In this paper, we explore how DHTs and mobile ad-hoc networks (MANETs) fit together. We argue that both share key characteristics in terms of self organization, decentralization, redundancy requirements, and limited infrastructure. However, node mobility and the continually changing physical topology pose a special challenge to scalability and the design of a DHT for mobile ad-hoc networks. In this paper, we show that with some local knowledge we can build a scalable and mobile structured peer-to-peer network, called mobile hash table (MHT). Furthermore, we argue that with little global knowledge, such as a map of the city or whatever area the nodes move in, one can even further improve the scalability and reduce DHT maintenance overhead significantly, allowing MHT to scale up to several ten thousands of nodes

In this paper, the authors introduced topology-based distributed hash tables (T-DHT) as an infrastructure for data-centric storage, information processing, and routing in ad hoc and sensor networks. T-DHTs do not rely on location information and work even in the presence of voids in the network. Using a virtual coordinate system, a distributed hash table which is strongly oriented to the underlying network topology was constructed. Thus, adjacent areas in the hash table commonly have a direct link in the network. Routing in the T-DHT guarantees reachability and introduces low hop-overhead compared with the shortest path.