sponsorship: VLAIO|HBC.2024.0406

The latency gap between wired and wireless networks poses a challenge in the adoption of wireless technologies in latency-sensitive scenarios. The gap is especially notable in multi-hop communication typical for industrial sensor networks and robotic swarms. The main reason behind it is that commonly used wireless protocols rely on store-and-forward routing and costly overhead procedures to avoid interference. This article introduces RF-Zero-Wire, an RF-based symbol-synchronous communication protocol. Instead of relaying the whole frame per hop in a store-and-forward manner, nodes concurrently relay the frame symbol by symbol, without the need for tight time synchronization. Based on data collected in real-world experiments, we reveal that the inevitable carrier frequency offsets (CFOs) introduced by imperfect crystal oscillators cause a beating effect under concurrent symbol transmissions. This is characterized by periodic constructive and destructive interference, which significantly affects reliability. Subsequently, a thorough simulation study shows how the beating problem can be overcome with error correction codes. RF-Zero-Wire allows achieving an end-to-end latency of less than 1 ms for a small 4-byte frame transmitted across 5 hops. Moreover, latency is shown to increase only by 0.16% per extra hop for 16-byte frames, which is negligible compared to the over 100% per-hop latency increase observed in store-and-forward protocols. The trade-offs between network reliability and CFO range, communication distance, node density, and achievable data rate are studied in large-scale experiments based on simulation.

Although federated learning offers a certain degree of privacy by aggregating user gradients instead of raw data, it remains vulnerable to various attacks, such as model poisoning. Existing defense mechanisms often address poisoning threats at the cost of exposing gradient information, which can lead to privacy risks such as member inference attacks. While techniques like cryptography or differential privacy can be employed to mitigate these risks, they often come with significant efficiency trade-offs. At the same time, a non-IID heterogeneous environment is also a big challenge. To address these challenges holistically, this paper proposes a dual-layer detection scheme (EDDFL). It combines norm-based filtering and isolation forest detection to effectively filter out malicious gradients, thereby preserving model accuracy even in adversarial environments. Furthermore, we incorporate a gradient quantization method that not only protects gradient privacy but also improves communication efficiency. Compared with existing approaches, the proposed method effectively addresses the challenges of model poisoning, gradient leakage, and data heterogeneity under non-IID settings. Experimental results demonstrate that our scheme significantly reduces both computational and communication overhead while maintaining privacy guarantees.

Swarm robotics focuses on designing and coordinating large groups of relatively simple robots to perform tasks in a decentralised and collective manner. The swarm provides a resilient and flexible solution for many applications. However, contemporary swarm robots have a significant power problem in that secondary (i.e. rechargeable) batteries are slow to charge and offer lifetimes of only a few years, increasing maintenance costs and pollution due to battery replacement. We imagine a different future, wherein battery-free robots powered by supercapacitors can be recharged in seconds, offer long-life autonomous operation and can rapidly pass charge between one another using trophallaxis. In pursuit of this vision, we contribute the CapBot, a battery-free swarm robot equipped with Mecanum wheels, a Cortex M4F application processor and Bluetooth Low Energy networking. The CapBot fully recharges in 16 s, offers 51 min of autonomous operation at top speed, and can transfer up to 50 % of its available charge to a peer via trophallaxis in under 20 s. The CapBot is fully open source and all software and hardware source is available online.

Battery-Powered Internet of Things (IoT) devices are widely deployed in applications like smart cities, industrial automation, healthcare, and agriculture. Concurrently, expanded cellular coverage provides ubiquitous connectivity, enabling transmission of diverse sensor data. However, conventional 4G and 5G devices designed for general use consume high power, limiting battery life to just a few days. Although cellular IoT networks such as LTE-M and NB-IoT offer lower power consumption, generic and reusable platforms remain scarce. This paper addresses this gap by introducing the Real-Time Actuator and Sensor Platform (RTASP), a software/hardware platform for large-scale cellular sensing supporting low-latency sensor and actuator streams, edge intelligence, and low-power operation. We provide an open-source hardware and software implementation and evaluate RTASP in a real-world industrial scenario, demonstrating low-latency multimedia streaming (initiating remote streams within seconds) and significantly longer battery life compared to conventional platforms (standby time of several months). These features enable the deployment of long-life sensors that uniquely support complex multimedia sensing and processing workflows.

Battery-powered IoT devices incur significant maintenance cost and complexity in remote and large-scale environmental monitoring deployments due to regular battery replacement. Energy harvesting in combination with supercapacitor charge storage offers a promising solution, but practical deployments are few and far between. This paper contributes to application research in this space by presenting an experience report on deploying a battery-free sensing system for wireless monitoring of forest microclimates. Our approach combines adaptive power management hardware and intelligent scheduling algorithms, which are compared against a baseline rechargeable battery-based solar alternative. In both cases, the wireless devices are used to drive a sensing system that reports atmospheric pressure, temperature, humidity, air quality and light levels over the LoRaWAN network. This data is used by environmental scientists to study and better understand the forest ecosystem. We evaluate our battery-free approach against the battery-powered baseline in a two-month forest monitoring deployment, demonstrating reliable operation using only harvested energy, while highlighting important lessons learned for researchers and practitioners deploying battery-free sensing systems.

Most Internet of Things (IoT) devices rely on batteries with operational lifetimes of only a few years, resulting in high maintenance costs and significant environmental impact from battery waste. Energy harvesting (EH) combined with supercapacitor-based energy storage has emerged as a promising alternative, enabling longer-lived and more sustainable IoT systems. However, the design and deployment of such battery-free systems remain challenging due to their complexity and perceived unreliability. To overcome these barriers, we present Flute, a plug-and-play (PnP), battery-free IoT platform designed for remote sensing applications. Flute integrates energy harvesting, supercapacitor storage, and NB-IoT connectivity to enable autonomous long-term operation. In this paper, we demonstrate Flute’s capabilities through a contactless waterlevel sensing application, highlighting its potential to simplify the development and deployment of sustainable IoT systems.

The developers of critical networked applications currently relied on kernel-level protections to ensure the timely delivery of critical messages like Wireless Emergency Alert (WEA) or multi-factor authentication (MFA) notifications. However, the increasing complexity of mobile operating system kernels and network stacks increases the attack surface for adversaries to exploit. In this paper, we introduce a system which safeguards network availability for critical mobile applications against powerful attackers. We achieve this by using the Trusted Execution Environment (TEE) found in most mobile devices to host minimal network drivers. Further, we utilize Trusted I/O to ensure that critical messages reach the end-user even if the device’s kernel is compromised. To demonstrate the feasibility of our approach, we provide a PoC implementation that mimics multi-factor authentication. Our Evaluation demonstrates that latency for all applications is reduced by around $21 \%$ on a representative mobile platform (ARM Cortex A9), though a significant throughput performance is observed.

Body area networking is an important segment of the wireless landscape, connecting a growing range of devices such as fitness trackers and smart watches using short range wireless links. However, conventional wireless technologies such as BLE and ANT provide poor support for low-latency applications such as haptic feedback due to their dependence upon duty-cycling to maintain acceptable power consumption. Recent physical-layer design efforts introduce a low-power alternative by establishing capacitively coupled communication through the human body. This mitigates duty-cycling requirements, but results in channel access times similar to those found in more conventional wireless networks, failing to provide adequate latency guarantees even for modestly sized networks. We address this problem by introducing ENOCH, an ultra-low power 12 MHz receiver that relies on capacitive coupling to establish a fieldbus-like through-body network. ENOCH delivers a sub-μW receiver, 10 kbps link-layer throughput, traffic prioritisation, 97% baseline reliability and 7 ms latency in a 7-node network, regardless of contention for the channel. Our design is simple and uses only commercial off-the-shelf components. As it requires neither large antennas nor crystal oscillators, we believe that further miniaturization beyond the current 1 cm2hardware/software prototype is possible.

Acoustic Anomaly Detection (AAD) has gained significant attention for the detection of suspicious activities or faults. Contrastive learning-based unsupervised AAD has outperformed traditional models on academic datasets, however, its model training is predominantly based on datasets containing only normal samples. In real industrial settings, a dataset of normal samples can still be corrupted by abnormal samples. Handling such noisy data is a crucial challenge, yet it remains largely unsolved. To address this issue, this paper proposes a Selective Contrastive learning framework Against Noisy data (SCAN) to mitigate the adverse effects of training the AAD model with anomaly-corrupted data. Specifically, SCAN progressively constructs confidence sample pairs based on the Mahalanobis distance, which is derived from the geometric median. These selected pairs are then integrated into the contrastive learning framework to enhance representation learning and model robustness. Extensive experiments under varying levels of label noise (i.e., the proportion of mislabeled abnormal samples in training data) demonstrate that SCAN outperforms state-of-the-art (SOTA) AAD methods on the real-world industrial datasets DCASE2022 and DCASE2024 Task2.

Today’s wireless sensor networks focus on achieving reliable data transfer over a lossy medium at the expense of latency. However, sensor data are often noisy and thus only lossily characterise real-world phenomena, rendering their exact transfer wasteful. Furthermore, many next-generation privacy-sensitive applications, such as smart grid control, real-time distributed object tracking, and inter-vehicle federated learning face latency and traffic bottlenecks due to the sheer amount of data collection required to overcome noise. We tackle this problem by introducing Twofer, a communication approach which reduces latency and traffic in high-noise or high-privacy settings by abandoning the focus on reliable networking. Twofer empowers developers to tune networks for latency-bound rather than reliability-bound performance; the system coordinates ambiguous transmissions, which are used to estimate the network-wide distribution of data, rather than to reliably communicate exact data from individual nodes. Twofer’s full-stack design maintains black-box compatibility with existing application code, but advocates for, and shows the value of, uncommon physical-layer features such as symbol-synchronous communication. The system is therefore implemented and evaluated on a prototype low-latency wireless mesh network called Zero-Wire. Experiments using state-of-the-art local differential privacy protocols show 25–75% latency reductions relative to conventional approaches. The results are also future-proof, with performance advantages increasing with the strength of the privacy guarantees that are offered.

Recommender systems are one of the most successful applications of using AI for providing personalized e-services to customers. However, data sparsity is presenting enormous challenges that are hindering the further development of advanced recommender systems. Although cross-domain recommendation partly overcomes data sparsity by transferring knowledge from a source domain with relatively dense data to augment data in the target domain, the current methods do not handle heterogeneous data very well. For example, using today’s cross-domain transfer learning schemes with data comprising clicks, ratings, user reviews, item metadata, and knowledge graphs will likely result in a poorly performing model. User preferences will not be comprehensively profiled, and accurate recommendations will not be generated. To solve these three challenges—handling heterogeneous data, avoiding negative transfer, and dealing with data sparsity—we designed a new end-to-end deep A dversarial M ulti-channel T ransfer network for C ross- D omain R ecommendation named AMT-CDR . Heterogeneous data is handled by constructing a cross-domain graph based on real-world knowledge graphs—we used Freebase and YAGO. Negative transfer is prevented through an adversarial learning strategy that maintains consistency across the different data channels. Data sparsity is addressed with an end-to-end neural network that considers data across multiple channels and generates accurate recommendations by leveraging knowledge from both the source and target domains. Extensive experiments on three dual-target cross-domain recommendation tasks demonstrate the superiority of AMT-CDR compared to eight state-of-the-art methods. All source code is available at https://github.com/bjtu-lucas-nlp/AMT-CDR .

A growing range of networked embedded devices are moving away from batteries and towards super-capacitor charge storage. However, mobile robots remain largely dependent upon batteries with slow recharge cycles and limited lifetimes. In this demonstration paper, we introduce a novel battery-free platform for swarm robotics which features: 24 minutes of operation running at its top speed of 1.24 km/h, a carrying capacity of over 2.5kg, full recharge cycles of under 12 seconds and rapid peer-to-peer charge transfer or trophallaxis in the field. This is supported by an nRF52840 Cortex-M4F equipped with BLE/ANT/802.15.4 transceiver. Notably, while the autonomy of FreeBots is limited compared to battery-powered robots, their operational vs charging duty-cycle is significantly higher at over 99%.

Memory corruption vulnerabilities remain a prevalent threat on low-cost bare-metal devices. Fuzzing is a popular technique for automatically discovering such vulnerabilities. However, bare-metal devices lack even basic security mechanisms such as Memory Management Unit. Consequently, fuzzing approaches encounter silent memory corruptions with no visible effects, making even discovery difficult. Once discovered, it is also essential to identify the type of observed vulnerability for applying mitigation. Both discovery and identification remain open challenges in the case of fuzzing firmware binaries. This article addresses these problems by proposing an automated instrumentation technique that allows the observation of memory corruption vulnerabilities that are otherwise not observable and facilitates the automated identification of the observed vulnerability. Additionally, we surveyed state-of-the-art IoT fuzzers and analyzed their experimental methodologies. We found that existing approaches have fundamental problems that lead to incorrect or misleading results. To evaluate the effectiveness of IoT fuzzers, it is essential to determine the range and type of vulnerabilities that these fuzzers can discover. Thus, we propose the first ground-truth benchmark suite for IoT fuzzers that enables accurate and consistent evaluation of their vulnerability-finding performance. Our instrumentation framework's efficacy and efficiency in combination with state-of-the-art IoT fuzzers are assessed using the proposed benchmark.

Low-power wide-area networks enable large-scale deployments of low-power wireless devices. LoRaWAN is a long-range wireless technology that has emerged as a low-power and low data rate solution to support Internet of Things applications. Although LoRaWAN provides a low-power and cost-efficient networking solution, recent literature shows that it performs poorly in terms of reliability and security in dense deployments due to the uncoordinated (ALOHA-based) nature of the MAC (medium access control) protocol. Furthermore, LoRaWAN is not robust against selective jamming attacks. This article proposes CRAM: a time-synchronized cryptographic frequency hopping MAC protocol designed for the LoRa physical layer. CRAM reduces the contention by fairly exploiting the available frequency space and maximizes the entropy of the channel hopping algorithm. We develop a large physical testbed and a simulator to thoroughly evaluate the proposed protocol. Our evaluations show that CRAM significantly improves reliability and scalability and increases channel utilization while making selective jamming difficult to perform compared to the standard LoRaWAN protocol.

Low Power Wide Area Networks (LPWANs) provide the Internet of Things (IoT) with energy-efficient, low cost, and long-range networking. LoRaWAN is one of the most widely deployed and studied LPWAN technologies due to its license-free operation and open standard, which has enabled 3rd parties to deploy millions of networks worldwide. LoRa allows transmission range and speed to be traded off by configuring the Spreading Factor (SF) setting. By tuning this parameter, the performance envelope of the network can be configured to favor a higher speed or longer range. While the SF of LoRA provides a simple and elegant means to tailor performance, at the highest spreading factor, maximum throughput drops to a few bits per second, dramatically increasing data extraction times. In this paper, we address this problem by proposing Smart-Hop, a novel multi-hop MAC layer protocol for LoRa which implements low-overhead routing. In contrast to prior work, Smart-Hop exploits the variable range afforded by different SFs to eliminate much of the complexity and overhead of traditional multi-hop mesh networking. Smart-Hop reduces the number of end-devices operating at high SFs while implementing low overhead routing in order to reduce data extraction times. These improvements are achieved while preserving the maximum range offered by LoRa in its highest SF configuration.

Intelligent Environments (IEs) enrich the physical world by connecting it to software applications in order to increase user comfort, safety and efficiency. IEs are often supported by wireless networks of smart sensors and actuators, which offer multi-year battery life within small packages. However, existing radio mesh networks suffer from high latency, which precludes their use in many user interface systems such as real-time speech, touch or positioning. While recent advances in optical networks promise low end-to-end latency through symbol-synchronous transmission, current approaches are power hungry and therefore cannot be battery powered. We tackle this problem by introducing BoboLink, a mesh network that delivers low-power and low-latency optical networking through a combination of symbol-synchronous transmission and a novel wake-up technology. BoboLink delivers mesh-wide wake-up in 1.13ms, with a quiescent power consumption of 237µW. This enables building-wide human computer interfaces to be seamlessly delivered using wireless mesh networks for the first time.

Faults in industrial equipment lead to significant costs due to down-time and unplanned maintenance interventions. Acoustic Internet of Things (IoT) sensors combined with machine learning offers the possibility of early fault detection to mitigate these costs. How-ever, prior approaches to Acoustic Anomaly Detection (AAD) are poorly suited to operating within the resource constraints of IoT systems as they require the acquisition of large volumes of data to train models from scratch for different machine types or oper-ating environments. To overcome the limitations, we introduce a system utilizing pretrained low-dimensional features and Gaussian mixture models. Preliminary results on real-world datasets show that our proposed approach outperforms state-of-the-art solutions based on the area under the curve (AUC) score with an average of 4 different machine types. Furthermore, our approach requires far less training data, making it more suitable to operate within the power and network constraints of IoT devices.

Sensor networks gather data at one or more gateway devices by sequentially receiving data frames from individual sensor nodes. In cases where applications rely on data that is distributed across sev-eral nodes, the latency of such node-by-node data collection scales with the number of nodes and the frame sizes involved, impeding fast decision-making. This paper seeks to overcome that limitation for latency-sensitive classification problems by introducing One-Take, a cross-layer aggregation protocol that aims to decouple the process of data gathering from network size and frame length. The protocol leverages dominance properties between concurrently transmitted symbols to convey socalled collages, messages that de-scribe relationships between data held by several nodes, rather than local information from any single node. Collages thus transmit a hash of distributed data that is computed within the physical layer, thereby communicating more informative features than regular frames of the same length. Experiments on a recent wireless mesh network platform, called Zero-wire, demonstrate that two-byte collages are delivered correctly 99% of the time and, using public sensor network data sets, show that One-take achieves a latency reduction of circa 42% relative to node-by-node transmissions.

The short lifespan of conventional smartphone batteries leads to toxic waste and increases replacement costs. In addition, contemporary devices require reliable electrical infrastructure which remains unavailable in major parts of the developing world. To address these problems, we propose a BAttery-free Mobile Interactive device (BaMbI), which aims to offer a scaled down set of smartphone features. BaMbI is based around an ARM Cortex M33 module with integrated LTE-M. In place of a battery, BaMbI uses a solar panel and an array of super-capacitors that offer sustainable operation and that can be fully charged in under one minute when mains power becomes available.

The Internet of Things (IoT) is increasingly connecting the most intimate parts of our daily lives to the Internet via connected ecosystems of hardware and software products. However, how these ecosystems operate and, in particular their impact on user privacy remains an open question. In this paper, we explore this question by analysing IFTTT, a popular task automation platform based on trigger-action programming. Through IFTTT, end users can easily create applets, aka recipes, that glue different IoT devices and online services together. While IFTTT brings many online services together, such as social media and mobile applications, its use for IoT automation is appealing for many users. Hence, analyzing the IFTTT ecosystem over time, enables us to study IoT trends and adaptation rate in real-user settings. In this paper we describe the IFTTT evolution by analyzing two existing IFTTT data sets and compiling a recent data set of its services and applets. This analysis exposes new platform characteristics and trends. Our data set, which was collected in October 2021, contains data about 694 services, 50898 applets, and 7003 endpoints. It also contains data about the information exchange in the platform via ingredients. In addition, we identify applets that might impose privacy risks for their users by detecting sensitive information flows in the system. We find that almost 30% of installations in the platform involve applets that utilize privacy-sensitive information such as users’ personal information, location, and health data. This trend is consistently present in the three analyzed data sets and, disturbingly, increases over time.

With the emergence of mobile devices having enough resources to execute real-time ML inference, deployment opportunities arise on mobile devices while keeping privacy-sensitive data close to the source and reducing server load. Moreover, offloading inference to a cloud server facilitates deployment of neural network-based applications on resource-constrained devices. Depending on the application goals and execution context of the application, the optimal deployment on either cloud server or mobile device varies during the lifetime of an application. In this paper, we propose a context-aware middleware that enables optimization of deployed application software to satisfy the application’s functional goals in accordance with changing execution context and environmental conditions. We facilitate system design through the abstraction of deployed software components as states and make use of finite state machines and contextual triggers to model the reconfiguration of the system. We evaluate our framework using a real-world nutritional monitoring application via food image recognition deployed in a two-tier mobile and cloud architecture. We compare the proposed solution with various static deployments of the application and show that our approach can react to changing application goals at run-time in order to reduce server load and thereby increase scalability.

Internet of Things (IoT) is becoming integrated into nearly every aspect of our modern life. Indeed, exploitation of such devices can directly lead to physical consequences in the real world. Previous work has shown that IoT devices can be compromised by exploits in lower software layers such as the Operating System (OS). Embedded Trusted Execution Environments (TEEs) provide a small Trusted Computing Base (TCB) to protect sensitive codes and data in such devices. TEEs assume a strong threat model where even a privileged attacker (e.g. OS) cannot compromise the confidentiality and integrity of the execution. Nevertheless, it has been shown that side channel attacks make it challenging to keep secrets during application execution. Interrupt latency side channel attacks (a.k.a. Nemesis) are a novel type of timing attacks that target embedded TEEs and extract application secrets from them. Nemesis attacks exploit the CPU’s interrupt mechanism to reveal microarchitectural instruction timings from embedded TEEs. Specifically, the attacker measures the latency of a precisely timed interrupt to differentiate between secret-dependent branches. In this paper, we present NemesisGuard, the first mitigation mechanism against such side channel attacks that does not require a modified compiler or hardware and can protect COTS binaries without access to source code. NemesisGuard applies a novel static binary instrumentation technique to balance secret-dependent branches in IoT application binaries. Evaluation of NemesisGuard shows that it mitigates Nemesis side channel attacks effectively and efficiently.

Battery-free Internet-of-Things devices equipped with energy harvesting hold the promise of extended operational lifetime, reduced maintenance costs, and lower environmental impact. Despite this clear potential, it remains complex to develop applications that deliver sustainable operation in the face of variable energy availability and dynamic energy demands. This article aims to reduce this complexity by introducing AsTAR, an energy-aware task scheduler that automatically adapts task execution rates to match available environmental energy. AsTAR enables the developer to prioritize tasks based upon their importance, energy consumption, or a weighted combination thereof. In contrast to prior approaches, AsTAR is autonomous and self-adaptive, requiring no a priori modeling of the environment or hardware platforms. We evaluate AsTAR based on its capability to efficiently deliver sustainable operation for multiple tasks on heterogeneous platforms under dynamic environmental conditions. Our evaluation shows that (1) comparing to conventional approaches, AsTAR guarantees Sustainability by maintaining a user-defined optimum level of charge, and (2) AsTAR reacts quickly to environmental and platform changes, and achieves Efficiency by allocating all the surplus resources following the developer-specified task priorities. (3) Last, the benefits of AsTAR are achieved with minimal performance overhead in terms of memory, computation, and energy.

This paper starts from the observation that mobile and edge devices are powerful enough to execute Machine Learning (ML) application components, which in turn creates opportunities to keep privacy-sensitive data close to its source. Composing and deploying a distributed ML application is far from trivial because the optimal configuration depends on the application’s goals and execution context, both of which may change throughout its lifetime. Prior research on context-aware reconfigurations in ML based applications offer limited capabilities for dynamically migrating software components between mobile, edge and cloud devices. In this paper, we propose a context-aware middleware that enables automated optimizations of the application deployment in order to satisfy the application’s functional goals while the execution context changes in terms of available computation, memory and network resources. We use finite state machines to model the reconfiguration of the application based on contextual triggers and facilitate system design through the abstraction of system states. We illustrate the benefits of our approach with an image recognition application with well-defined performance goals that is deployed in a three-tier mobile-edge-cloud architecture.

The Internet of Things (IoT) can transform any object into a smart thing by means of sensors that measure physical properties, which in turn send messages wirelessly over the Internet. Many IoT applications are time-sensitive, i.e. they react to events based on their time context, ranging from industrial control to mobility automation business cases. When sensor messages fail to arrive in a timely fashion, time-sensitive applications can react to timeouts by replacing missing measurements with predictions to maintain timely responses, albeit less accurate. To produce meaningful results, these applications combine measurements based on their time context; a challenging task since sensor messages can be late or lost due to the unreliability and varying delay of wireless IoT systems. Whereas timely measurements can be combined to obtain new results, late measurements can be used to recompute past results that were partially based on predictions, thus improving their accuracy over time. The reactive programming paradigm is a good fit for IoT application development, because reacting to events from data streams very well embodies the intuition behind sensors in a wireless sensor network. However, existing reacting programming frameworks offer limited support for reacting to events based on their time context, which remains a repetitive process that is tedious and error-prone, forcing developers to perform manual checks that result in cross-cutting code fragments. To fill this gap, this paper contributes a suite of programming language concepts to combine and operate on events based on their time context, with which we extend ReactiveX, a state-of-the-art reactive programming framework. We evaluate our extensions by implementing a swarm robotics use case with and without them, and comparing the resulting programs in terms of code size. The results show that using our extensions decreases the resulting code size by 77.7%, thus simplifying the development of time-sensitive IoT applications using reactive programming.

Modern streaming analytics platforms, such as Apache Flink, can receive, process, and act on real-time streaming data from various sources to take immediate action as events occur in the real world. Combined with the emergence of the Internet of Things (IoT), these platforms have revolutionized many application domains by deriving strategic insights from incoming sensor data. For instance, manufacturers can retrofit production lines with sensors, enabling them to rapidly detect and correct problems by analyzing incoming sensor data in real-time. However, when sensors generate data too slowly, the application produces incorrect results due to insufficient information. Conversely, generating data too fast rapidly decreases sensor battery lifetime and wastes storage resources. Consequently, the data generation rate of the IoT infrastructure must match the data processing rate of the application(s) at all times, which is a challenging task due to the several facets of IoT dynamism: sensor mobility, sensor failure, changing application data requirements, varying message inter-generation and transmission delay. Currently, application developers are required to manually configure sensor sampling periods to match these rates; a time-consuming process that is inefficient, if not ineffective, due to the dynamism of the IoT infrastructure. In this paper, we tackle this problem by introducing Ermis, a novel middleware that runs on the gateway and automatically adapts sensor sampling periods at runtime to match application requirements, without relying on any previous knowledge about the underlying IoT infrastructure. To the best of our knowledge, this is the first solution that can ensure the presence of sufficient data for the IoT applications under various sources of dynamism, while minimizing unnecessary transmissions to extend sensor battery lifetime. Extensive evaluation on real-time data from a physical testbed demonstrated a ~34% increase in the expected battery lifetime, while consistently ensuring data sufficiency, compared to state-of-the-art.

Vibration monitoring uses data gathered from accelerometers to study kinetic phenomena in applications such as: structural health monitoring and predictive maintenance. The Internet of Things (IoT) has the potential to greatly expand the range and scope of vibration monitoring applications by delivering long-life wireless sensors that can be cost-effectively embedded in hard to reach places such as; within machines, infrastructure or the built environment. However, achieving this vision is difficult due to the stringent resource constraints of contemporary IoT devices and networks. This has led the research community to develop a creative range of application-specific near-sensor processing firmware. However, systematic support for generic vibration monitoring on resource-poor IoT networks remains an open problem. We tackle this challenge by introducing ReFrAEN, a software framework that efficiently enables a wide range of vibration monitoring applications on IoT networks. ReFrAEN achieves this through a deeply configurable combination of compression techniques and data processing algorithms. These features allow end-users to effectively trade-off between resource consumption and data resolution in order to meet battery life constraints while preserving sufficient data quality to support the target application. Our evaluation shows that ReFrAEN is capable of identifying bearing faults, while dramatically improving battery lifetime and reducing latency in comparison to prior approaches.

Recent innovations in energy harvesting promise extended operational life and reduced maintenance costs for the next generation of Internet of Things (IoT) platforms. However, energy management in these platforms remains problematic due to dynamism in energy supply and demand, inefficiency in storing and converting energy and a lack of per-task charge isolation. This paper tackles this problem by proposing a software defined charge storage module called Morphy, which combines a polymorphic capacitor array with intelligent power management software. Morphy delivers energy to application tasks in a flexible, efficient, and isolated manner. Morphy provides two software extensions to the Operating System scheduler: the energy semaphore blocks the execution of tasks until sufficient charge is available to safely run them, and the energy watchdog monitors and mitigates energy management bugs. We have realized a prototype of Morphy with the hardware form factor of a standard 9V (PP3) battery package and a software library that integrates with the FreeRTOS scheduler. Our evaluation shows that, in comparison to standard energy storage and management approaches, our prototype reaches an operational voltage more quickly, sustains operation longer in the case of power failure and effectively isolates charge storage for dedicated tasks with minimal compute, memory and energy overhead.

Low Power Wide Area Network (LPWAN) technologies offer the advantage of wide coverage areas and low power consumption for low data-rate Internet-of-Things (IoT) applications. LoRaWAN, the Long Range Wide Area Network is a key technology in this space, with a growing worldwide presence. LoRa devices are expected to operate autonomously for extended periods in order to support diverse IoT applications. Despite being energy efficient, frequent battery replacements are typically required over the lifetime of a LoRa device. This increases maintenance costs and furthermore, disposing of large numbers of dead batteries is damaging to the environment. Energy harvesting offers a potential solution, but it is difficult to ensure sustainability. In this paper, we propose EH-CRAM, a centralised Kalman filter-based optimisation algorithm where the gateway is responsible for controlling End-Device configurations (i.e: data transmission rates, spreading factors and energy harvesting period) based upon incoming traffic and solar energy, thus balancing energy supply and demand. In addition, by using a time-synchronised cryptographic frequency hopping scheme, EH-CRAM also tackles the issues of energy efficiency and performance. Our evaluation shows that EH-CRAM significantly reduces contention, while maximising reliability and energy efficiency to support sustainable energy-harvesting for LoRa EndDevices (ED's).

Accurate indoor positioning remains an open research question. Existing solutions are either expensive, short-range, or inaccurate. A new approach is therefore required to cost-effectively and accurately support localization in large indoor environments. We tackle this problem by introducing a practical optical localization scheme, called OSLo, that cost-effectively scales to support large buildings. OSLo uses a meshed network of low-cost cameras as localization anchors and smart LEDs as tags, which transmit their IDs and context sensor data over the meshed cameras using optical communication. OSLo is capable of localizing dense deployments of tags with an accuracy of under 1 meter at a distance of 35 meters from the localization anchor. Furthermore, smart LED tags can be manufactured for less than $1. We systematically evaluate the performance of OSLo in the context of a real-world car localization use-case at Ford Motor Company in Germany and demonstrate promising results in terms of detection distance, and localization accuracy.

The Internet of Things (IoT) is being deployed in an ever-growing range of applications, from industrial monitoring to smart buildings to wearable devices. Each of these applications has specific computational requirements arising from their networking, system security, and edge analytics functionality. This diversity in requirements motivates the need for adaptable end-devices, which can be re-configured and re-used throughout their lifetime to handle computation-intensive tasks without sacrificing battery lifetime. To tackle this problem, this article presents Chimera, a low-power platform for research and experimentation with reconfigurable hardware for the IoT end-devices. Chimera achieves flexibility and re-usability through an architecture based on a Flash Field Programmable Gate Array (FPGA) with a reconfigurable software stack that enables over-the-air hardware and software evolution at runtime. This adaptability enables low-cost hardware/software upgrades on the end-devices and an increased ability to handle computationally-intensive tasks. This article describes the design of the Chimera hardware platform and software stack, evaluates it through three application scenarios, and reviews the factors that have thus far prevented FPGAs from being utilized in IoT end-devices.

Low power wide area networks (LPWANs) are being applied in many Internet of Things applications around the globe. These technologies offer economic coverage of wide areas, while retaining low power operation. LoRaWAN is a key technology in this space, with a world-wide presence and millions of devices deployed in the field. Despite this early success, recent research has shown that LoRa performs poorly in dense deployments with a high degree of contention. Furthermore, LoRa is not robust against selective jamming attacks. In this paper, we propose CRAM: a cryptographic frequency hopping MAC protocol designed for the LoRa physical layer that reduces contention by fairly exploiting all available frequency space, while making it significantly more difficult to perform selective jamming. Our evaluation shows that CRAM significantly reduces contention, thereby dramatically increasing scalability and reliability in comparison to the standard LoRa protocol.

The Internet of Things (IoT) is being deployed at large scale in a wide range of long-life applications. Examples range from Industry 4.0 to smart lighting systems. These applications have diverse requirements of non-volatile storage. However, the flash memory that is used in today's IoT devices offers limited write endurance and must therefore be carefully managed if applications are to deliver on their promises of multiyear lifetimes. Managing the health of flash memory is difficult for application developers, as it requires in-depth hardware and software knowledge, which often needs to the problem being neglected. While various techniques have been proposed to preserve the health of flash memory, prior work tends to focus on a single hardware platform and data type. Furthermore, prior work does not provide lifetime guarantees. This paper tackles this problem by proposing MicroVault, a simple and unified interface for reliable non-volatile data storage on resource-constrained IoT devices. MicroVault enforces developer-specified lifetime guarantees through a range of lifetime extension techniques, which are adaptively applied based upon the needs of the application. Evaluation shows that MicroVault dramatically extends the lifetime of flash memory while minimising overhead.

The Internet of Things (IoT) consists of embedded sensors that transmit events over a wireless network. Complex event processing provides powerful abstractions to aggregate and analyze relationships among event streams in real time, which can improve IoT application development and management. A key challenge are stream imperfections, because of the non-deterministic nature of IoT: events can be delayed due to variance in delay, or may even be missing due to packet loss. Timeouts are used to handle stream imperfections, distinguishing between delayed and lost events. For some applications, missing delayed events is costly, as the quality of the result depends on the presence of all inputs, but reacting too late to event non-arrivals can also lead to incorrect results. State-of-the-art results in timeouts that are impractically large when little to no missed events are tolerated by the application. We propose Kairos, a novel, self-configuring technique for determining event arrival timeouts in IoT that are both small and ensure that little to no events are missed, while also eliminating the overhead and complexity of user configuration. We evaluate our approach against the state-of-the-art using two representative IoT networks: SmartMesh IP and LoRaWAN. The results show that our solution is capable of reducing timeouts by up to two orders of magnitude even when little to no missed events are tolerated, thus satisfying the aforementioned application requirements.

The performance dichotomy between wired and wireless networks for the Internet of Things primarily arises from the inherent complexity and inefficiency of networking abstractions such as routing, medium access control and store-and-forward packet switching. This paper aims to enable a new class of latency-sensitive applications by breaking all three of these abstractions to deliver a performance envelope that resembles that of a wired bus in terms of deterministic latency and throughput. The essence of this approach is a novel networking paradigm for optical wireless communication, referred to as a symbol-synchronous bus, wherein a mesh of nodes concurrently transmit LED-based signals. This paper realises the paradigm within a platform called Zero-Wire and evaluates it on a 25-node testbed under laboratory conditions. Key end-to-end performance measurements on this physical prototype include 19 kbps of contention-agnostic goodput, interface-level latency under 1 ms for two-byte frames across four hops, jitter on the order of 10s of μs, and a base reliability of 99%. These first results indicate a bright future for the under-explored area of optical wireless mesh networks in delivering ubiquitous connectivity through a simple and low-cost physical layer.

Despite the ubiquitous deployment and development of wireless technology for the Internet of Things (IoT), contemporary radio frequency (RF)-based solutions still cannot match the performance of a "wire" in terms of latency and throughput. This abstract presents a demonstration of Zero-Wire, a novel optical wireless approach that addresses this gap to enable latency-sensitive IoT applications. The essence of this approach is a new networking paradigm, referred to as a symbol-synchronous bus, wherein a mesh of nodes concurrently transmits optical signals. The demonstration setup is composed of 25 Zero-Wire nodes, forming a mesh network, and the demo showcases the network's behavior during a series of transmissions. End-to-end performance measurements include 19 kbps of contention-agnostic goodput, latency under 1 ms for two-byte frames, jitter on the order of 10s of μs, and a base reliability of 99%.

Application development for Cyber Physical Systems (CPS) is challenging, because the wireless network and the devices introduce latencies that vary continuously along with the load, status, or environmental conditions of the infrastructure. Reactive programming is well suited for the development of event-driven applications, yet current reactive programming frameworks require developers to predict event arrival time-boundaries at compile time, which is impractical, if not impossible, for CPS. Thus, there is a tradeoff between timeliness and completeness of complex event computations, e.g., operational efficiency in a manufacturing plant: Waiting too long until all individual events arrive can fail to produce a useful result, while not waiting long enough may lead to faults due to incomplete status information. In this article, we propose (a) a set of extensions to state-of-the-art reactive programming frameworks, which remove the burden of specifying timeouts at compile time by utilizing (b) Khronos, a middleware that automatically determines timeouts by taking into account variations in event arrival times due to the underlying infrastructure. Evaluation on a physical testbed shows that the extensions significantly decrease developer effort and that Khronos considerably improves timeliness under varying network configurations and conditions, while still satisfying the application’s tolerance to missed events.

Applications for the Internet of Things (IoT) are often data-centric. Data-centric routing then enables messages to reach relevant consumers while avoiding flooding and explicit resource discovery. This reduces the amount of transmissions required to support relevant applications and provides energy savings as well as a convenient programming abstraction: messages can be addressed to nodes that advertise features matching a constraint. In low-power wireless mesh networks, such feature-oriented routing traditionally relies on costly and inflexible network overlays. Recent work establishes lightweight support for diverse data-centric traffic patterns, but sacrifices expressiveness of feature-oriented functionality and hence applicability. This paper overcomes that trade-off by introducing SMRFET, a multicasting system that integrates data-centric functionality into a standard low-power network stack. SMRFET thus improves over the art: it offers more expressive addressing mechanics (range queries over a node's features) at lower implementation and runtime cost (no additional networking mechanisms). Additionally, SMRFET can be configured to handle memory constraints: its performance degrades gracefully as the designated amount of memory decreases. SMRFET therefore brings lightweight and expressive group communication to wireless IoT networks.

Recommender systems have gained prominence in bringing users tailor-made content from the web aiding their decision making process. However, this personalization comes at a cost of privacy of sharing personal information with the recommendation provider. Moreover, with growing sizes of datasets and models, centralized processing of such data has become challenging. To this end, we propose a federated matrix factorization algorithm to enable personal data to be stored and used on-device for training while sending updates to train a centralized model. We illustrate preliminary results from our algorithm applied to recommendation of articles to users of a mobile application based on their reading history. We compare our performance with centralized matrix factorization applied on the same dataset.

Applications for the Internet of Things (IoT)are often data-centric. Data-centric routing then enables messages to reach relevant consumers while avoiding flooding and explicit resource discovery. This kind of routing thus provides energy savings as well as a convenient programming abstraction: messages can be addressed to nodes that advertise features matching a constraint. In low-power wireless mesh networks, such feature-oriented routing traditionally relies on costly and inflexible network overlays. Recent work establishes lightweight support for diverse data-centric traffic patterns, but sacrifices expressiveness of feature-oriented functionality and hence applicability. It is also unclear whether the energy consumption advantages offered by data-centric routing extend to this new lightweight approach. To address these concerns, this paper introduces the SMRFET system. SMRFET improves the expressiveness of state-of-the-art feature-oriented routing by supporting numeric rather than binary features. The system integrates data-centric functionality into group-based multicast and thus adds only a small amount of overhead: experiments show that SMRFET significantly reduces the required amount of message passing relative to alternative systems for group communication. Additionally, SMRFET can be reconfigured to handle memory constraints: its performance degrades gracefully as the amount of memory allocated to it decreases. SMRFET therefore brings lightweight and expressive group communication to the wireless IoT.

The Internet of Things (IoT) is shaped by the increasing number of low-cost Internet-connected embedded devices that are becoming ubiquitous in every aspect of modern life. With their cost-sensitive design, integrating hardware-based security mechanisms into such devices is undesirable. Therefore, securing these devices is a particularly difficult challenge, especially, due to their growing popularity as attack targets, via remote malware infestations. The vast majority of such devices are bare-metal, where they execute programs in fully-accessible and unprotected memories without any operating system and even without including any form of security. This is beside the fact that IoToperating systems offer little or no protection. This paper addresses this problem through the concept of a Security MicroVisor (SμV), which provides embedded devices that lack hardware-based memory protection units with memory isolation using software virtualisation and assembly-level code verification. More specifically, our contribution is two-fold. First, we propose SμV as a software-based memory isolation technique. We then formally verify the software architecture, written in C, to prove that it is memory-safe and crash-free. Second, we propose a software-based remote attestation, as an example of a fundamental security service that can be implemented on top of SμV, to detect malware-infected devices. We first describe the design and implementation of SμV. Then, we highlight the formal verification of software architecture, and characterize the remote attestation protocol. We evaluate the SμV implementation using an 8-bit AVR microcontroller that is widely used in IoT devices. Evaluation results show that SμV provides strong security guarantees while maintaining extremely low overhead in terms of memory footprint, performance, and power consumption. Furthermore, we extend the performance evaluation also to the remote attestation scheme, illustrating its limited overhead.

The Internet of Things (IoT) consists of embedded devices that sense and manage our environment in a growing range of applications. Large-scale IoT systems such as smart cities require significant investment in both equipment and personnel. To maximize return on investment, IoT platforms should support multiple third-party applications and adaptation of infrastructure over time. Realizing the vision of shared IoT platforms demands strong security guarantees. That is particularly challenging considering the limited capability and resource constraints of many IoT devices.

The Internet of Things (IoT) promises to tackle a range of environmental challenges and deliver large efficiency gains in industry by embedding computational intelligence, sensing and control in our physical environment. Multiple independent parties are increasingly seeking to leverage shared IoT infrastructure, using a similar model to the cloud, and thus require constrained IoT devices to become microservice-hosting platforms that can securely and concurrently execute their code and interoperate. This vision demands that heterogeneous services, peripherals and platforms are provided with an expanded set of security guarantees to prevent third-party services from hijacking the platform, resource-level access control and accounting, and strong isolation between running processes to prevent unauthorized access to third-party services and data. This paper introduces Polyglot CerberOS, a resource-secure operating system for multi-tenant IoT devices that is realised through a reconfigurable virtual machine which can simultaneously execute interoperable services, written in different languages. We evaluate Polyglot CerberOS on IETF Class-1 devices running both Java and C services. The results show that interoperability and strong security guarantees for multilingual services on multi-tenant commodity IoT devices are feasible, in terms of performance and memory overhead, and transparent for developers.

Long range, low power networks are rapidly gaining acceptance in the Internet of Things (IoT) due to their ability to economically support long-range sensing and control applications while providing multi-year battery life. LoRa is a key example of this new class of network and is being deployed at large scale in several countries worldwide. As these networks move out of the lab and into the real world, they expose a large cyber-physical attack surface. Securing these networks is therefore both critical and urgent. This paper highlights security issues in LoRa and LoRaWAN that arise due to the choice of a robust but slow modulation type in the protocol. We exploit these issues to develop a suite of practical attacks based around selective jamming. These attacks are conducted and evaluated using commodity hardware. The paper concludes by suggesting a range of countermeasures that can be used to mitigate the attacks.

The widespread use of mobile devices has allowed the development of participatory sensing systems that capture various types of data using the existing sensors on mobile devices in order to upload the data to cloud based services for later use. Gathering data from such sources requires a mechanism to establish trust on the sensor data. For example an application may require a proof of authenticity of sensor readings originating from anonymous sources. Establishment of trust on the sensor data has been addressed in the literature. However, in many cases this sensor data needs to be preprocessed on the device itself before being uploaded to the target server. This processing could include resizing of images, hiding identifiable faces and sensitive data in images, anonymization of GPS data etc. while ensuring the chain of trust from data capture to the delivery of data to the consumer. There is a need for a framework that provides a means to implement arbitrary operations to be performed on trusted sensor data while guaranteeing the authenticity of the data. This paper presents the design and implementation of a framework that allows the capture of trusted sensor data, the development of trusted operations on sensor data, and provides a mechanism for performing predefined trusted operations on the sensor data such that the chain of trust is maintained. Evaluation shows that the performance of the proposed system is reasonable and that the trust guarantees are strong.

The state-of-practice for Internet of Things (IoT) applications is deployment on specialised networks of embedded devices connected to a cloud backend. While this paradigm has successfully supported a range of IoT systems, its power is limited by the high latency and bandwidth caused by communications with remote data servers and the inability to share specialised IoT infrastructure across applications. To improve these aspects, this paper proposes re-imagining all resources of the IoT infrastructure as microservice-hosting platforms. Applications decomposed as a set of services can then share IoT resources and run communicating modules closer together, tightening control loops and reducing latency and communications. To this end, we present Niflheim, a generic end-to-end middleware that provides modular microservice-based orchestration of applications to deploy and manage them on all resources across the tiers of the IoT, from IoT end-devices through gateways to the cloud. This enables increased flexibility in deployment and operations, while remaining efficient in terms of hardware and software requirements. We evaluate Niflheim in a smart building use case and demonstrate improved latency and bandwidth consumption for applications, while enabling efficient shared use of the IoT infrastructure resources.

The state-of-practice for Internet of Things (IoT) applications is deployment on specialised networks of embedded devices connected to a cloud backend. This paradigm is limited by the high latency and bandwidth incurred by communications with remote data servers and the inability to share specialised IoT infrastructure across applications. Efficiency can be improved by re-imagining all resources of the IoT infrastructure as micro-service hosting platforms. Applications decomposed as a set of services can then share IoT resources and run communicating modules closer together, tightening control loops and reducing latency and communications. This demo showcases Niflheim, a generic end-to-end middleware that provides modular microservice-based orchestration of applications on all resources across the tiers of the IoT, from IoT end-devices through gateways to the cloud. We demonstrate that this enables increased flexibility in application deployment and operations, while remaining efficient in terms of hardware and software requirements.

Due to its scale and dynamism, the Internet of Things (IoT) requires efficient and flexible communication support. At the network layer, IPv6 integrates heterogeneous technologies to provide interoperability, efficient multicast group communication and a flexible address space. At the application layer, publish-subscribe (pub-sub) middleware implements a scalable, dynamic and loosely-coupled data dissemination scheme. The pub-sub paradigm is a natural use case for IPv6 multicast but the two mechanisms are poorly integrated in the IoT. We tackle this problem by proposing a framework that integrates pub-sub middleware and multicast to reduce communication overhead. Our solution maps application-layer subscriber groups to network-layer multicast groups. Pub-sub hosts can either implicitly derive the necessary multicast address or request it from a group manager. We evaluate our framework on an IoT network testbed composed of representative hardware and demonstrate improvements in bandwidth and energy consumption that scale with the size of the network. Bandwidth consumption of a publishing sensor decreases by up to 54% for 10 subscribers and 66% for 20 subscribers. Moreover, the implementation has a minimal memory footprint, requiring only an additional 1.3% dynamic memory and 4.7% flash storage.

Creating, deploying and configuring applications for Internet of Things (IoT) scenarios today remains complex and costly for many users. The MicroPnP platform addresses this complexity problem and provides a true zero-configuration and standards-based solution that radically reduces the cost of acquiring, building, and operating wireless sensing and actuation IoT systems at scale. MicroPnP combines true Plug-and-Play integration of sensing and actuation peripherals with ultra-reliable wireless mesh networking and extreme battery lifetimes. MicroPnP was awarded in an international IoT competition, and is currently being successfully used in commercial IoT scenarios.

Internet of Things (IoT) applications require diverse sensors and actuators. However, contemporary IoT devices provide limited support for the integration of third-party peripherals. To tackle this problem, we introduce μPnP: a hardware and software solution for plug-and-play integration of embedded peripherals with IoT devices. μPnP provides support for: driver development, automatic integration of third-party peripherals, discovery and remote access to peripheral services. This is achieved through a low-cost hardware identification approach, a lightweight driver language and a multicast network architecture. Evaluation shows that μPnP has a minimal memory footprint, reduces development effort and provides true plug-and-play integration at orders of magnitude less energy than USB.

Developing reliable application for the Internet-of-Things (IoT) is challenging due to heterogenous sensors and the resource-constraints of IoT platforms. Multimedia sensors such as cameras and microphones require significant memory and bandwidth. Typically, multimedia content results in bursty traffic. It is shown in the literature that the bursty traffic degrades the performance of the network. Hence, the regulation of bursty transmission is important for achieving high reliability. In this paper, we propose Dawn, a dependable networking framework for time-synchronised IoT platforms, which uses application meta-data to derive bandwidth requirements and uses this information to optimally allocate bandwidth for the network. Dawn regulates the transmission of bursty traffic generated by multimedia sensors, while performing admission control for dynamically added sensors. Dawn guarantees end-to-end reliability for heterogenous IoT applications with minimal energy consumption. Our preliminary evaluation results show that dawn is dependable and provides 100% reliability for dynamic multimedia-enabled IoT applications while increasing the network lifetime by up to 4 years.

In the participatory sensing model, humans may serve as opportunistic sensors and flexible actuators while also consuming sensing services. Integrating humans into sensing systems has the potential to increase scale and reduce costs. However, contemporary participatory sensing software provides poor consideration of user dynamism, which includes: mobility across networks, mobility across devices and context-awareness. To address these limitations we propose the User Component and User Bindings. The former represents the user as a first class reconfigurable element of evolving and shared participatory sensing platforms. The latter allows the middleware to support multiple communications channels including Online Social Networks (OSN) to connect users with sensing applications. Our approach increases user participation, reduces out-of-context interactions and only consumes a limited amount of energy by sharing context information between applications. We support these claims by evaluating our approach on a two weeks experiment in which three participants take part in three concurrent participatory applications.

Sensor network deployments, such as smart cities and smart logistics, are evolving towards open infrastructure on top of which multiple parties deploy and make use of each other's application software. Of vital importance in such distributed settings, is an agreement on message semantics and contents. Yet, the current state-of-the-art in sensor programming typically remains low-level in terms of messaging, and lacks support for coordination across platforms and stakeholders. We present Talk Sens, a message definition framework that facilitates development of correctly interacting application logic by means of a shared data model, a message description language and serialisation code generation. Additionally, integration of Talk Sens with an existing component model results in run-time retrievable message definitions that facilitate third-party interactions. Our evaluation shows that message sizes, software sizes and middleware overhead remain well within the bounds of sensor network resource constraints.

Smart applications enable pervasive and informed interactions between the physical and digital worlds. These applications are deployed on resource constrained wireless sensor networks and are commonly implemented using software modularization schemes such as components and services. Composing smart applications at the abstraction level offered by embedded software modularization schemes is complex and time consuming. The complexity of composing them is derived from the need to understand many low-level issues, e.g. embedded programming languages, coordination mechanisms, software tool chains. We present an application composition service that reduces composition effort by offering a declarative specification of sense, process and consume flows. We demonstrate reduction in composition effort for three real-world smart applications deployed on a smart office environment.

Contemporary Wireless Sensor Networks like Smart Offices and Smart Cities are evolving to become multi-purpose application hosting platforms. These WSN platforms can simultaneously support multiple applications which may be managed by multiple actors. Reconfigurable component models have been shown to be viable solutions to reducing the complexity of managing and developing these applications while promoting software re-use. However, implicit parameter dependencies between components make reconfiguration complex and error-prone. Our approach achieves automatic composition-safe re-parametrization of distributed component compositions. To achieve this, we propose the use of language annotations that allow component developers to make these dependencies explicit and constraint-aware network protocols to ensure constraint propagation and enforcement.

Wireless Sensor Networks (WSN) are evolving beyond research prototypes towards real world deployments in various application domains. While prior research has resulted in a range of operating systems and associated programming languages, a comprehensive empirical analysis of WSN operating systems is missing from the literature. We address this problem through an empirical study of all actively maintained WSN operating systems for the popular Tmote Sky / TelosB platform: TinyOS, Contiki and Lorien. Our analysis considers overhead at each stage of the software life cycle. During the development phase, we measure developer effort in terms of lines of application code. During the execution phase we measure energy consumption, flash footprint and RAM usage. During the reconfiguration phase we measure artefact size and developer effort in terms of number of configuration commands. Our results indicate distinct trade-offs in terms of development effort, application performance and reconfiguration performance. We find that TinyOS performs best for static applications with tight RAM constraints, while Contiki offers the lowest development effort and Lorien performs best in dynamic applications which require reconfiguration.

Wireless Sensor Networks (WSNs) are subject to high levels of dynamism arising from changing environmental conditions and application requirements. Reconfiguration allows software functionality to be optimized for current environmental conditions and supports software evolution to meet variable application requirements. Contemporary software modularization approaches for WSNs allow for software evolution at various granularities; from monolithic re-flashing of OS and application functionality, through replacement of complete applications, to the reconfiguration of individual software components. As the nodes that compose a WSN must typically operate for long periods on a single battery charge, estimating the energy cost of software evolution is critical. This paper contributes a generic model for calculating the energy cost of the reconfiguration in WSN. We have embedded this model in the LooCI middleware, resulting in the first energy aware reconfigurable component model for sensor networks. We evaluate our approach using two real-world WSN applications and find that (i.) our model accurately predicts the energy cost of reconfiguration and (ii.) component-based reconfiguration has a high initial cost, but provides energy savings during software evolution.

Creating and managing applications for Wireless Sensor Networks (WSNs) is complicated by large scale, resource constraints and network dynamics. Reconfigurable component models minimize these complexities throughout the application lifecycle. However, contemporary component based middleware for WSNs is limited by its poor support for distribution. This paper introduces the Loosely-coupled Component Infrastructure (LooCI), a middleware for building distributed component-based WSN applications. LooCI advances the state-of-the-art by cleanly separating distributed concerns from component implementation, supporting application-level interoperability between heterogeneous WSN platforms and providing compatibility testing of bindings at runtime. Together, these features promote the safe and efficient composition and reconfiguration of distributed WSN applications. We evaluate the performance of LooCI on three classes of sensor nodes and demonstrate that these features can be provided with minimal overhead in terms of computation, memory and message passing.

SUMMARY Achieving a clean separation of concerns is a well known approach to improving system adaptability and evolution. We propose to apply this principle to run‐time reconfigurable component models for networked embedded systems. By separating configuration properties from runtime component instances, we achieve: (i) improved support for concurrent component use, (ii) optimized resource use, and (iii) reduced effort in runtime configuration management. We demonstrate how this approach can be seamlessly implemented on existing component models for wireless sensor networks without imposing additional constraints and without changes to their API or coordination model. Furthermore, significant memory savings are achieved in concurrent scenarios. Copyright © 2012 John Wiley & Sons, Ltd.

Wireless Sensor Networks are increasingly being deployed in enterprise scenarios involving multiple actors. The capabilities of sensors must be shared across many applications and sensors must cooperate across federations spanning administrative domains. This paper describes our efforts towards constructing federated wireless sensor systems. It leverages and details key building blocks designed to achieve federation and sharing: a component model with distributed event-based communication, a policy-driven control infrastructure for resources and communications, a deployment architecture and a mechanism to establish and propagate trust into the wireless sensor network. The building blocks focus on the specific and necessary in-network extensions, such as extensions to sensor run-time, middleware and programming abstractions, rather than the backend challenges. The combination of the building blocks creates a security middleware that supports multiparty federated sensor networks. An estimate of the total cost of federation in terms of footprint is provided and first experiences of application of the middleware are reported upon.

Considerable research has been performed in applying run-time reconfigurable component models to the domain of wireless sensor networks. The ability to dynamically deploy and reconfigure software components has clear advantages in sensor network deployments, which are typically large in scale and expected to operate for long periods in the face of node mobility, dynamic environmental conditions and changing application requirements. To date, research on component and binding models for sensor networks has primarily focused on the development of specialized component models that are optimized for use in resource-constrained environments. However, current approaches impose significant overhead upon developers and tend to use inflexible binding models based on remote procedure calls. To address these concerns, we introduce a novel component and binding model for networked embedded systems (LooCI). LooCI components are designed to impose minimal additional overhead on developers. Furthermore, LooCI components use a novel event-based binding model that allows developers to model rich component interactions, while providing support for easy interception, re-wiring and re-use. A prototype implementation of our component and binding model has been realised for the SunSPOT platform. Our preliminary evaluation shows that LooCI has an acceptable memory footprint and imposes minimal overhead on developers.

In order to provide an increasing number of functionalities and benefit from sophisticated and application-tailored services from the network, distributed applications are led to integrate an ever-widening range of networking technologies. As these applications become more complex, this requirement for 'network heterogeneity' is becoming a crucial issue in their development. Although progress has been made in the networking community in addressing such needs through the development of network overlays, we claim in this paper that the middleware community has been slow to integrate these advances into middleware architectures, and, hence, to provide the foundational bedrock for heterogeneous distributed applications. In response, we propose our 'open overlays' framework. This framework, which is part of a wider middleware architecture, accommodates 'overlay plug-ins', allows physical nodes to support multiple overlays, supports the stacking of overlays to create composite protocols, and adopts a declarative approach to configurable deployment and dynamic reconfigurability. The framework has been in development for a number of years and supports an extensive range of overlay plug-ins including popular protocols such as Chord and Pastry. We report on our experiences with the open overlays framework, evaluate it in detail, and illustrate its application in a detailed case study of network heterogeneity.

Engineering adaptive software is an increasingly complex task. Here, we demonstrate Genie, a tool that supports the modelling, generation, and operation of highly reconfigurable, component-based systems. We showcase how Genie is used in two case-studies: i) the development and operation of an adaptive flood warning system, and ii) a service discovery application. In this context, adaptation is enabled by the Gridkit reflective middleware platform.

Abstract Flooding is a growing problem, which affects more than 10% of the U.K. population. The cost of damage caused by flooding correlates closely with the warning time given before a flood event, making flood monitoring and prediction critical to minimizing the cost of flood damage. This paper describes a wireless sensor network (WSN) for flood warning, which is capable of not only integrating with remote fixed‐network grids for computationally intensive flood modelling purposes but also performing on‐site grid computation. This functionality is supported by the reflective and component‐based GridKit middleware, which provides support for both WSN and grid application domains. Copyright © 2007 John Wiley & Sons, Ltd.

This paper describes the open P2P tracing system which aims to improve the research community's understanding of P2P file sharing systems by providing continuous and up-to-date traffic data which is anonymized and made freely accessible to all interested parties. It is our hope that this open data set will grow over time into a resource capable of exposing trends in P2P network usage and promote research into the socio-technical factors that drive user behaviour on P2P file sharing systems.

Current approaches to flood monitoring (e.g. in river valleys) involve statically deploying depth and ultrasoundbased flow sensors across flood-prone areas, and feeding the collected data off-site (e.g. using GSM) to grid-based

The cost of damage caused by flooding is directly related to the warning-time given before a flood occurs. Therefore, improving the coverage, accuracy and reliability of flood prediction systems is of great importance. This paper proposes a novel Grid-based approach to supporting flood prediction through the use of embedded sensor nodes equipped with wireless networking technology. These nodes implement a light-weight Grid capable of collecting and transmitting data gathered by flood sensors for off-site analysis. Additionally, the nodes are capable of performing basic on-site analysis that is used to inform system behavior. The proposed system utilizes Lancaster's next-generation component-based Grid middleware platform 'GridKit', which offers inherent support for reconfiguration, embedded devices and heterogeneous network technologies, making it ideally suited to this application.

There has been a significant body of research conducted into various structured and unstructured overlay network protocols. Both paradigms have advantages for specific application domains and researchers are beginning to examine the benefits of using hybridized systems. We hypothesize that resource awareness and adaptation are essential to the efficient exploitation of the resources available on the diverse nodes which compose peer-to-peer networks. To support this, we propose a hybrid peer-to-peer model which uses an unstructured decentralised network layered on top of a structured overlay to provide support for multiple levels of adaptation.

Thorough testing of distributed systems, particularly peer-to-peer systems can prove difficult due to the problems inherent in deploying, controlling and monitoring many nodes simultaneously. This problem will only increase as the scale of distributed systems continues to grow. This framework implements a test bed environment using a semi-centralized peer-to-peer network as a substrate for sharing resources made available from standard PCs. This framework automates the process of test-case deployment using a combination of reflection and aspect oriented programming. This allows 'point-and-click' publishing of software onto the test-bed. Our framework also provides a common monitoring, control and logging interface for all nodes running on the network. Together, these features greatly reduce deployment-time for real-world test scenarios. Automated insertion and removal of test code also ensures that the testing process does not compromise the correctness of the final system.

The first generation of peer-to-peer file sharing systems followed the traditional client-server paradigm. However, legality and scalability issues have driven the development of decentralized file sharing protocols; the most popular of these being Gnutella. To date, such systems have been unable to match the quality of service (QoS) offered by centralized architectures. AGnuS improves QoS on Gnutella by increasing file availability, improving network friendliness and increasing file quality. This is achieved by layering caching, load balancing, content-based routing and filtering services on top of the core Gnutella protocol.