Low power sensing networks often operate in open and potentially hostile environments. Ensuring that only legitimate devices communicate with the network is paramount. Authentication serves as the first line of defense in securing communication and data integrity, thereby protecting the network and devices from unauthorized access and data breaches. However, the ease of access to devices and sensors in the Internet-Of-Things (IoT) makes it easier for an attacker to replace legitimate devices and implant rogue ones in their stead; or physically tamper with devices (e.g., by moving them to another location). In this paper, we propose a resilient machine learning approach to uniquely identify deployment environments based on the link quality footprints of the devices that transmit from these environments. Our approach complements device identification based on unique RF transmission footprints. To the best of our knowledge, this is the first approach that attempts to uniquely identify the deployment environment despite considerable variations in both external and internal factors that affect signal propagation. We employ two different machine learning models, one based on a Convolutional Neural Network (CNN) and the other based on a Residual Network (ResNet). Through independent experiments involving actual deployments in five different environments in Miami, Florida (land, lake, Biscayne Bay, South Beach, and Crandon Beach), we attest that both models were able to uniquely identify the deployment environments with an average accuracy exceeding 99%. Furthermore, our models were able to distinguish between specific deployment configurations. In general, the ResNet model correctly identified the type of prototypes used with 100% accuracy (and CNN, with 98% accuracy). Both models were able to identify the types of radio used for transmission with 99% accuracy.

Abstract A large number of heterogeneous wireless networks share the unlicensed spectrum designated as the ISM (Industry, Scientific, and Medicine) radio band. These networks do not adhere to a common medium access rule and differ in their specifications considerably. As a result, when concurrently active, they cause cross-technology interference (CTI) on each other. The effect of this interference is not reciprocal, the networks using high transmission power and advanced transmission schemes often causing disproportionate disruptions to those with modest communication and computation resources. CTI corrupts packets, incurs packet retransmission cost, introduces end-to-end latency and jitter, and make networks unpredictable. The purpose of this paper is to closely examine its impact on low-power networks which are based on the IEEE 802.15.4 standard. It discusses latest developments on CTI detection, coexistence and avoidance mechanisms as well on messaging schemes which attempt to enable heterogeneous networks directly communicate with one another to coordinate packet transmission and channel assignment. Graphical abstract

Emerging autonomous and semi-autonomous vehicles such as Unmanned Aerial Vehicles (UAV) and Unmanned Surface Vessels (USV) use unlicensed frequency bands to interact with their remote control stations and peers. Typically, these systems are cost-intensive and their safe operation is of paramount importance. As a result, they establish powerful wireless communication links. When low-power IoT sensing nodes operate nearby, their performance can be considerably affected by cross-technology interference arising from the powerful systems. Different coexistence strategies have been proposed to deal with cross-technology interference, including dynamic channel-hopping, low-level spectrum sensing and channel adaptation, channel blacklisting, and direct cross-technology communication. These approaches require advanced spectrum scanning, detection, and clustering as well as knowledge of low-level packet structure and modulation schemes. In this paper, we propose a packet transmission strategy relying on link quality statistics alone to deal with cross-technology interference. Our approach does not require intimate knowledge of modulation schemes; nor does it require the modification of any hardware components. Evaluation based on traces of field experiments show that our approach improves Packet Delivery Ratio (PDR) by more than 30% when compared to baseline results and by more than 20% when compared to state-of-the-art solutions.

In the past two decades, several applications have been proposed for Wireless Sensor Networks (WSNs). Some of these applications, such as habitat monitoring, active volcano monitoring, and toxic gas detection, exclude the involvement of human presence because the environments are either dangerous, inaccessible, or too extensive. The scope of these applications can be extended if the ground networks are assisted by a network of Unmanned Aerial Vehicles (UAVs). Paramount to this is the stability of the wireless links the UAVs establish with the ground nodes. Lossy and unstable links are costly to maintain the UAVs may exhaust their batteries prematurely. In this paper, we experimentally investigate the stability of aerial-to-ground links and propose a stochastic model to predict link quality as a function of time. The model can be used to estimate goodput, determine mission duration, and estimate short-term link connection and disconnection durations.

The future, in which mobile robots and human beings intermingle in industrial complexes, shopping malls, airports, and similar areas, is not far. The condition, however, requires the realization of several features, including self-localization, self-navigation, identification and avoidance of obstacles, and dynamic route discovery. Indoor localization has been the subject of interest for over two decades, most recent advances attempting to take advantage of freely available signals from a plethora of indoor sources. As far as the estimation task is concerned, most of the models employ extended kalman filters and particle filters. Each technique has its own merits and demerits, as well as a set of assumptions. The purpose of this paper to identify the most significant components of these techniques and to closely examine the prevailing assumptions underlying the selection process of indoor localization techniques. Moreover, the paper experimentally demonstrates the error modeling process in kalman and particle filters.

Wireless sensor networks accommodating the mobility of nodes will play important roles in the future. In residential, rehabilitation, and clinical settings, sensor nodes can be attached to the body of a patient for long-term and uninterrupted monitoring of vital biomedical signals. Likewise, in industrial settings, workers as well as mobile robots can carry sensor nodes to augment their perception and to seamlessly interact with their environments. Nevertheless, such applications require reliable communications as well as high throughput. Considering the primary design goals of the sensing platforms (low-power, affordable cost, large-scale deployment, longevity, operating in the ISM band), maintaining reliable links is a formidable challenge. This challenge can partially be alleviated if the nature of link quality fluctuation can be known or estimated on time. Indeed, higher-level protocols such as handover and routing protocols rely on knowledge of link quality fluctuation to seamlessly transfer communication to alternative routes when the quality of existing routes deteriorates. In this article, we present the result of extensive experimental study to characterise link quality fluctuation in mobile environments. The study focuses on slow movements (<5 km h -1 ) signifying the movement of people and robots and transceivers complying to the IEEE 802.15.4 specification. Hence, we deployed mobile robots that interact with strategically placed stationary relay nodes. Our study considered different types of link quality characterisation metrics that provide complementary and useful insights. To demonstrate the usefulness of our experiments and observations, we implemented a link quality estimation technique using a Kalman Filter. To set up the model, we employed two link quality metrics along with the statistics we established during our experiments. The article will compare the performance of four proposed approaches with ours.

A certain class of wireless devices such as electrocardiogram (ECG), electromyogram (EMG) and electroencephalogram (EEG), are very useful for telemedicine because they enable the free movement of patients while vital biophysical measurements are taken from them. However, these devices are very sensitive to motion artifacts - electric potentials generated due to the undesirable movement of electrodes on the surface of the skin or the change in the skin impedance. In this paper we examine the scope and usefulness of different types of model-based signal processing and dimensionality reduction techniques to model and reason about motion artifacts. While the techniques we review are applicable for a wide range of signals, we limit our analysis, nevertheless, to wireless electrocardiograms, so that we can base our investigation on experimental data.

Wireless electrocardiograms (WECG) facilitate the long-term monitoring of patients in their residential environment. However, the freedom of movement provokes motion artifacts in the measurements of the useful signals, which significantly affect the quality of the data. In this paper, we propose a tensor decomposition method to combine data from heterogeneous sources and remove motion artifacts. We transformed synchronously sampled electrocardiogram and inertial sensors into the time-frequency space using wavelet decomposition. Afterward, we formed a three-way tensor consisting of a single lead WECG and a motion reference. Thus we recorded measurements from eleven healthy subjects undertaking different types of movements, namely, Standing up, Bending forward, Walking, Running, Jumping, and Climbing stairs. An additional WECG sensor was attached at the back of each subject to measure motion with negligible cardiac input. This signal was subsequently added to a noise-free WECG segment to generate artificially corrupted signal. We factorize the measurement sets using Canonical Polyadic Decomposition to determine mutual information present in both sensor types (WECG and inertial sensor) and extract the motion artifacts from the noisy WECG. We evaluated the results by considering the Signal-to-Noise-Ratio and the Root Mean Squared Error between the actual and estimated artifacts.

The worldwide workload of the public cloud infrastructure has been increasing steadily for the past many years and the latest statistics indicate that it will remain so for the coming years. At the same time, the energy consumption of the cloud infrastructure is considerably high. In this respect, the efficient utilisation of computing resources is of profound importance. In contemporary data centres tens of thousands of virtual machines execute simultaneously. Considering the number and heterogeneity of the virtual machines, balancing the demand for and the supply of resources is one of the challenges facing Cloud and Edge Computing. Often, infrastructure providers over-provision resources to ensure that service level agreements are respected. This, however, is not sustainable and its long-term impact on the environment cannot be overlooked. In this paper, we propose the use of tensor decomposition to analyse the resource utilisation metrics of a large number of hosted virtual machines and to identify complementary and contentious features which can be vital for efficient resource utilisation.

Wireless electrocardiograms can be useful for a wide range of applications including early detection of seizures onset, West Syndrome, sleep apnoea, Temporal Lobe Epilepsy (TLE), supra-ventricular tachycardia, atrial flutter, and atrial fibrillation. One of the advantages of using wireless sensing platforms in telemedicine is that patients can freely move and carry out everyday activities unhindered whilst vital cardiac action potentials are being measured. However, the measurements are highly sensitive to motion as motion changes the electrical characteristics of the interface between the electrodes and the skin and, thereby, distorts the useful signal. In this paper we propose a strategy for establishing the statistics of motion artefacts. Our approach employs 3D accelerometers to reason about the movement affecting the electrodes of a wireless electrocardiogram and takes advantage of the structure and sequence of cardiac action potentials.

Wireless sensor networks are useful for a large number of healthcare applications which require mobile nodes. Most existing applications rely on body area networks (BAN) which require the presence of additional devices, such as mobile phones, to transfer data from the BAN to a remote base station or a server. In this paper we propose to extend BAN with personal area networks (PAN) so that enhanced mobility and seamless collection of data can be possible. In order to improve the reliability of this merge and support a high goodput, we also propose a seamless handover mechanism which enables mobile transmitters to discover and transfer communication to reliable relay nodes when the quality of an existing link deteriorates. In this paper we shall report how we implemented our scheme for TinyOS and TelosB platforms and compared it with four other competitive schemes.

Wireless sensor networks supporting the free mobility of nodes can be useful for several applications. For example, in residential areas and rehab centres, sensors can be attached to subjects to monitor their movements, body exertions, and cardiac activities. These benefits, however, are also challenged by the difficulty of establishing reliable and stable links. In cellular networks, mobile stations are always associated with the nearest base station through intra-and inter-cellular handover. The underlying process is that the quality of an established link is continually evaluated and handover decisions are dully made by resource rich base stations. In wireless sensor networks, should a seamless handover be carried out, the task has to be accomplished by energy constraint, resource-limited, and low-power wireless sensor nodes in a distributed manner. In this paper we propose a sender-initiated mobility management protocol to enable seamless handover. We have fully implemented the protocol in TinyOS environment for the TelosB and Imote2 platforms, experiment results showing that our protocol achieves high reliability and triggers less handover requests (less than 50% to 80%) compared to three state-of-the-arts. Furthermore, our protocol reduces the signalling overhead by up to 95%.

One of the strategies employed to deal with resource inefficiency in data centres is dynamic virtual machine/container consolidation. The idea behind is, by populating physical servers with an optimal number of virtual machines, all the server's resources (CPU, memory, network bandwidth, etc.) can be utilised effectively. This approach requires (1) the free migration of virtual machine at runtime and (2) the identification of virtual machines which exhibit complementary features. Most existing or proposed approaches are based on elaborate and complex multi-variate optimisation and do not easily lend themselves to fast and intuitive solutions. In this paper, we investigate the scope and usefulness of dimensionality reduction techniques, ideas borrowed from unsupervised machine learning, to analyse the existence of contentious and complementary features in the resource consumption characteristics of co-located virtual machines. Initial results suggest that fast and tractable scheduling can be achieved using these techniques.

Data centres and similar server clusters consume a large amount of energy. However, not all consumed energy produces useful work. Servers consume a disproportional amount of energy when they are idle, underutilised, or overloaded. The effect of these conditions can be minimised by attempting to balance the demand for and the supply of resources through a careful prediction of future workloads and their efficient consolidation. In this paper we extend the cutting stock problem for consolidating workloads having stochastic characteristics. Hence, we employ the aggregate probability density function of co-located and simultaneously executing services to establish valid patterns. A valid pattern is one yielding an overall resource utilisation below a set threshold. We tested the scope and usefulness of our approach on a 16-core server with 29 different benchmarks. The workloads of these benchmarks have been generated based on the CPU utilisation traces of 100 real-world virtual machines which we obtained from a Google data centre hosting more than 32000 virtual machines. Altogether, we considered 600 different consolidation scenarios during our experiment. We compared the performance of our approach-system overload probability, job completion time, and energy consumption-with four existing/proposed scheduling strategies. In each category, our approach incurred a modest penalty with respect to the best performing approach in that category, but overall resulted in a remarkable performance clearly demonstrating its capacity to achieve the best trade-off between resource consumption and performance.

Recently, bulk-data or burst transmission has been the focus of research in wireless sensor networks. The aim is to benefit applications which require high throughput but small scale deployment. Healthcare applications and applications which are intended to support independent living for the elderly are typical examples. The main idea is to exclusively provide the channel for one transmitter only until it has transferred all of the packets it has accumulated in the buffer. This exclusive use avoids aimless contention and significantly reduces packet transmission latency by dispensing with repeated clear channel assessment, random back-off, and the transmission of RTS and CTS packets for every single packet. Existing or proposed MAC protocols supporting bulk-data transmission, however, do not react well to link quality fluctuation, since nodes make repeated attempt to retransmit lost packets even when the statistics of received packets suggests that the channel is still bad or packet transmission will be deferred arbitrarily even though packet loss is an isolated and uncorrelated occurrence. In this paper, we address these issues and estimate the duration of good and bad links from the statistics of received ACK packets. Moreover, we provide a MAC layer solution to enable the coexistence of multiple transmitters during bulk-data transfer.

As the global IP traffic and its demand for computation increase in a rapid and sustained manner, processor, server, and network architectures are also undergoing a considerable evolution. Two of the manifestations of this evolution are the integration of a large number of computing nodes in a single server and the interconnection of many servers via high-speed communication links. At present, however, the node-to-node communication bandwidth is one of the severest resource bottlenecks in massively parallelized applications. There is a concerted effort by the academia and the industry to achieve higher data rate by assembling multiple parallel links. This effort, however, is inherently limited by many constrains, including space. Optical interconnects, on the other hand, promise superior data rates, lower transmission losses, and less inter-channel crosstalk when compared to electrical interconnects. Development in this area promise data rates in the range of Tera bits per second per link and beyond. So far, however, little attention is given to the power adaptiveness of optical interconnects. In this paper, we present an optical interconnect concept which adjusts its power consumption in response to the change in the statistics of the incoming workload. The several components of the link have been designed and developed in hardware. Based on initial power and performance measurements of the components, a link model of our optical interconnect was created. The performance-power consumption characteristics of this model was simulated applying different workload statistics and the potential of the energy savings by the adaptivity have been evaluated. It is revealed that the power consumption of our optical interconnect reduces by up to 40% when its workload was exponentially distributed (signifying underutilisation) compared to a Weibull distribution workload (signifying full capacity workload). This study confirms the high potential for power saving in performance adaptive optical interconnects.

The architectures of large-scale Internet servers are becoming more complex each year in order to store and process a large amount of Internet data (Big Data) as efficiently as possible. One of the consequences of this continually growing complexity is that individual servers consume a significant amount of data even when they are idle. In this paper we experimentally investigate the scope and usefulness of existing and proposed dynamic power management strategies to manage power at core, socket, and server levels. Our experiment involves four dynamic voltage and frequency scaling policies, three different workloads having different resource consumption statistics, and the activation and deactivation of different sockets (packets) of a multicore, multi-socket server. Moreover, we establish a quantitative relationships between the workload (w) and the estimated power consumption (p) under different power management strategies to make a quantitative comparison of the different strategies and server configurations.

Wireless sensor networks supporting aggressive sampling in harsh environments (to deal with high packet loss or to provide reliable data from sensors such as 3D accelerometers and 3D gyroscopes) require transmission schemes which can achieve a relatively high throughput. Existing contention-based, low-power listening MAC protocols are not apt for these types of networks because their channel utilisation is considerably low. In this paper we propose a hybrid burst transmission scheme to achieve high throughput between static relay nodes, such as nodes deployed on a civil infrastructure (bridge or building). Our transmission scheme deals with link quality fluctuations and adaptively adjust the number of packets that can be transmitted in burst. Its essential features are relying on statistics that are obtained (1) offline and reflect the long-term characteristic of a link and (2) online and reflect the short-term link quality fluctuation. We experimentally compared our transmission scheme with two proposed state-of-the-art schemes in terms of throughput, transmission delay, packet loss, and energy consumption. We implemented all transmission schemes and integrated them into the TinyOS environment and the TelosB platform.

Several reproducible experiments are required before actual deployment of wireless sensor networks takes place if stable and predictable outcomes of protocols and data processing algorithms are desired. Considering the typical size of wireless sensor networks and the number of parameters that can be configured or tuned, conducting repeated and reproducible experiments can be both time consuming and costly. The conventional way of evaluating the performance of different protocols and algorithms under different network configurations is by changing the source code and reprogramming the testbed, which requires some effort. In this paper, we propose a traffic flow control management system that facilitates the execution of repeated experiments in an efficient and flexible way. We implemented our system on top of TinyOS for the TelosB platform and demonstrated the scope and usefulness of the system by conducting several experiments in two real testbeds.

Understanding fluctuations of link quality in a wireless sensor network is useful for different reasons. For example, nodes can determine when and for how long they should transmit packets, so that they can reduce packet loss rate and the cost of retransmission (delay as well as power consumption). However, because the quality of a link depends on many factors, it cannot be known except in a probabilistic sense. In this paper we estimate the expected duration in which the quality of a specific link remains stable using the conditional distribution function of the signal-to-noise ratio (SNR) of received acknowledgment packets. We employ the expected duration to determine how long nodes should transmit packet in burst and how long they should refrain from contention. To develop our model, we deployed Imote2 sensor platforms in indoor and outdoor places and transmitted more than 70, 000 packets. We transmitted additional 16,900 packets to test our model. 90% of the time, our approach resulted in high packet delivery compared with the case in which packets were transmitted without knowledge of link quality fluctuations.

The energy consumption of a wireless sensor network affects its lifetime which in turn affects the scope and usefulness of the network. Most existing or proposed MAC protocols enable nodes to specify a duty cycle, so that they can sleep much of the time to save energy. However, only very few models exist to determine the appropriate time and duration of a sleep phase. Existing approaches rely on pre-calculated sleep durations or are difficult to implement on real platforms. We propose a runtime and adaptive model to estimate the sleep time and duration of wireless sensor nodes. Our model takes the statistics of incoming and outgoing packets at a relay node which is then supplied to a general queueing model. The model is lightweight and can be fitted into any existing MAC protocol. We have implemented our model for TelosB platform and TinyOS environment. We integrated our model with two existing protocols (TinyOS LPL MAC and XMAC) and compared the performance of these protocols with and without our model. The performance evaluation results show that the energy consumption of a relay node reduced by 11.4 - 64.8%. The overall throughput of the network increased by up to 24%. Moreover, our model readily responded to changes in packet traffic rate while at the same time increasing the packet transmission reliability by 64.5 - 67.4% for different traffic scenarios.

The amount of data that are computed, stored, and shared over the Internet is rising at an unprecedented scale. This has necessitated more servers to be deployed and drastic improvement in the capacity of individual servers. However, several independent studies also reveal that resources are not optimally utilised in most existing data centres and server clusters. Since the introduction of server virtualization and cloud computing, the research community has proposed several workload aggregation and dynamic consolidation techniques, however, most of these techniques are either theoretical and rely on simulation environments or use real servers but static workloads or benchmarks. In reality, the workload of data centres fluctuates as a function of time and servers frequently experience both overloading and underutilised conditions. In this paper we introduce the HAECubie demonstrator we developed and deployed to experimentally evaluate the scope and usefulness of dynamic workload consolidation in a server cluster and to quantitatively analyse the relationship between energy/power consumption and the utility (performance) that can be achieved through workload consolidation. Our demonstrator is a video hosting platform and enables Internet users to stream videos of variable length. The number of users accessing the HAECubie as well as the duration of videos they stream are modelled as stochastic processes based on realistic estimation of the workload of existing video hosting platforms.

This paper addresses link quality fluctuation and its impact on the packet delivery capacity of wireless sensor networks. Independent studies have previously confirmed that link quality fluctuates even in a static deployment and understanding stable durations, good and bad alike, can contribute to the efficient transmission of packets. We propose a two stage Markov model to characterise link quality fluctuation and to determine when and for how long nodes should transmit packets in burst. Both to develop and test our model, we deployed a wireless sensor network consisting of 14 nodes in a garden and transmitted more than 120,000 packets with different links. The experiment results confirm that our approach improved the packet delivery capacity of the links by up to 40% when compared with a baseline and by up to 25% when compared with a scheme that employs conditional distribution functions.

We experimentally investigate the mutual influence of application- and platform-level adaptations in a virtualized cluster environment. At the application level, applications can adapt to a changing execution environment by dynamically exchanging components that enable them to trade energy for utility and vice versa. Likewise, at the platform level, virtual machine monitors can migrate virtual machines from one server to another either to consolidate workloads and switch-off underutilized servers or to distribute the workload of overloaded servers. Our experiment quantify impacts of various types of adaptations on QoS, power consumption, and energy-overhead.

Several applications have been proposed for mobile wireless sensor networks. Some of these applications require the transfer of a large amount of data in a short period of time. This is challenging, since mobility can lead to a deterioration in the quality of an established link. Frequent link disconnection may in turn require a mobile node to repeatedly establish new links with the surrounding relay nodes to proceed with the data transfer. The new link establishment may cause extra data communication latency and make most of the applications delay sensitive. To evaluate the effect of mobility on latency, this paper first sets up a mathematical model based on a hybrid medium access control (MAC) protocol in mobile scenarios. It then uses NS2 simulation to further analyze the latency associated with mobility. Both results show that the latency increases with an increment in the network density and the duty cycle.

Quantitatively estimating the relationship between the workload and the corresponding power consumption of a multicore processor is an essential step towards achieving energy proportional computing. Most existing and proposed approaches use Performance Monitoring Counters (Hardware Monitoring Counters) for this task. In this paper we propose a complementary approach that employs the statistics of CPU utilization (workload) only. Hence, we model the workload and the power consumption of a multicore processor as random variables and exploit the monotonicity property of their distribution functions to establish a quantitative relationship between the random variables. We will show that for a single-core processor the relationship is best approximated by a quadratic function whereas for a dualcore processor, the relationship is best approximated by a linear function. We will demonstrate the plausibility of our approach by estimating the power consumption of both custom-made and standard benchmarks (namely, the SPEC power benchmark and the Apache benchmarking tool) for an Intel and AMD processors.

One of the mechanisms to achieve energy efficiency in virtualized/cloud environments is consolidation of workloads on an optimal number of servers and switching-off of idle or underutilized servers. Central to this approach is the migration of virtual machines at runtime. In this paper we investigate the cost (migration time) of virtual machines migration. We shall show that migration time exponentially increases as the available network bandwidth decreases; migration time linearly increases as the RAM size of a virtual machine increases. Furthermore, the power consumption of both the destination and the source servers remain by and large the same for a fixed network bandwidth, regardless of the VM size. Interestingly, for the same combination of virtual machines, different orders of migrations resulted in different migration time. We observed that migrating resource intensive virtual machines first yields the shortest migration time. In general, the migration time should be modeled as a random variable since the factors that affect it cannot be known except in a probabilistic sense. Therefore, we propose a probabilistic approach to quantify the cost of virtual machines migration.

We propose a MAC protocol that supports the mobility of nodes in wireless sensor networks. The protocol enables burst transmission and seamless handover to achieve high throughput and to reduce packet delivery latency and packet loss. An adaptive filter continuously evaluates the RSSI values of received acknowledgment packets and decides whether a mobile node should transfer a communication to a nearby relay node with a better link quality. The handover process itself takes place without breaking an existing link. This paper presents the design, implementation and evaluation of the MAC protocol.

The power consumption of presently available Internet servers and data centers is not proportional to the work they accomplish. The scientific community is attempting to address this problem in a number of ways, for example, by employing dynamic voltage and frequency scaling, selectively switching off idle or underutilized servers, and employing energy-aware task scheduling. Central to these approaches is the accurate estimation of the power consumption of the various subsystems of a server, particularly, the processor. We distinguish between power consumption measurement techniques and power consumption estimation models. The techniques refer to the art of instrumenting a system to measure its actual power consumption whereas the estimation models deal with indirect evidences (such as information pertaining to CPU utilization or events captured by hardware performance counters) to reason about the power consumption of a system under consideration. The paper provides a comprehensive survey of existing or proposed approaches to estimate the power consumption of single-core as well as multicore processors, virtual machines, and an entire server.

In this paper, we experimentally investigate the scope and usefulness of Dynamic Voltage and Frequency Scaling (DVFS) in multimedia servers. For our experiment, we considered four scaling policies, two heterogeneous servers, and two different application scenarios. In the first scenario, we used an IO-Intensive multimedia downloading application while in the second scenario we used a predominately CPU-Intensive application, namely, a video transcoder. We will show that while the advantage of DVFS for IO-Intensive applications is apparent, it is not so for CPU-Intensive applications. The advantage in IO-Intensive applications depends on the selection of frequencies in the machine and the way the CPU speed scales. We observed that the choice of a particular DFVS technique became more consequential when the machine had a wide selection of operation frequencies while a gradual change in the operation frequency was more energy efficient. For CPU-Intensive applications, the use of DVFS was counter productive and the overhead of scaling was not negligible.

Live migration, the process of moving a virtual machine (VM) interruption-free between physical hosts is a core concept in modern data centers. Power management strategies use live migration to consolidate services in a cluster environment and to switch off underutilized machines to save power. However, most migration models do not consider the energy cost of migration. This paper experimentally investigates the factors that affect the power consumption and the duration of virtual machine migration. We use the KVM platform for our experiment and show that a live migration entails an energy overhead and the size of this overhead varies with the size of the virtual machine and the available network bandwidth.

Several applications have been proposed for mobile wireless sensor networks. Some of these applications require the transfer of a large amount of data in a short period of time. This is challenging, since the mobile node may be required to repeatedly establish a link with multiple relay nodes which proceed to forward the data to the base station. Apart from the technical difficulty mobility may cause, there is an associated latency to the data transfer. One way to deal with the problem of latency is to timely foresee the deterioration in the quality of a link and to seamlessly transfer the communication to a more stable link. This paper extends the RI-MAC protocol to support a seamless handover. Once a mobile node realizes that its data packets cannot be completely transmitted before an existing link breaks, it will search for a new relay node without interrupting the communication with the current node. The paper sets up a mathematical model to investigate the latency associated with a handover. The analytic model quantifies the handover latency as a function of the network density and the duty cycle.

Dynamic voltage and frequency scaling (DVFS) has been a useful power management strategy in embedded systems, mobile devices, and wireless sensor networks. Recently, it has also been proposed for servers and data centers in conjunction with service consolidation and optimal resource-pool sizing. In this paper, we experimentally investigate the scope and usefulness of DVFS in a server environment. We set up a multimedia server which will be used in two different scenarios. In the first scenario, the server will host requests to download video files of known and available formats. In the second scenario, videos of unavailable formats can be accepted; in which case the server employs a trans coder to convert between AVI, MPEG and SLV formats before the videos are downloaded. The workload we generate has a uniform arrival rate and an exponentially distributed video size. We use four dynamic scaling policies which are widely used with existing mainstream Linux operating systems. Our observation is that while the gain of DVFS is clear in the first scenario (in which a predominantly IO-bound application is used), its use in the second scenario is rather counterproductive.

Most Internet applications employ some kind of load balancing policies in a cluster setting to achieve reliable service provision as well as to deal with a resource bottleneck. However, these policies may not ensure the utilization of \textit{all} of the hardware resources in a server equally efficiently. This paper experimentally investigates the relationship between the power consumption and resource utilization of a multimedia server cluster when different load balancing policies are used to distribute a workload. Our observations are the following: (1) A bottleneck on a single hardware resource can lead to a significant amount of underutilization of the entire system. (2) A ten times increment in the network bandwidth of the entire cluster can double the throughput of individual servers. The associated increment in power consumption of the individual servers is 1.2% only. (3) For TCP-based applications, session information is more useful than other types of status information to utilize power more efficiently. (4) The use of dynamic frequency scaling does not affect the overall throughput of IO-bound applications but reduces the power consumption of the servers; but this reduction is only 12% of the overall power consumption. More power can be saved by avoiding a resource bottleneck or through service consolidation.

Merging the features of Cloud computing, autonomic computing, pervasive computing, and mobile computing are now at its initial stage but the effort is visibly showing the benefits of these paradigms. A large number of applications can take advantage of this, including healthcare, traffic control, and social network applications. However, these applications are complex by nature and introduce several challenges of their own, for example, reliable sensing, accurate context recognition, scalability, security, and the challenge of dealing with previously unforeseen side-effects of adaptations. These challenges can be surmounted when researchers of diverse background come together and provide different views of the same problems and help each other understand the complex relationships between contending ideas. The Casemans 2012 workshop opens the necessary platform for researchers of ubiquitous computing, autonomic computing, and similar fields to address these issues.

In wireless sensor networks, saving energy is crucial in order to increase the network lifetime. Energy is often saved by synchronizing the nodes activity, and having long periods of inactivity, or by having nodes exchange a global activity schedule. The synchronization and the exchange of a global schedule are two examples where information is boadcast from a specific node to the whole network. In this paper, we focus on the delay required to broadcast information in the whole network using a tree topology. We first show that the diffusion delay can be significantly reduced by utilizing the parallelization of node processing. We provide an algorithm in order to find optimal solutions when transmissions are sequential. Then, we propose a linear algorithm that is able to find good solutions. We compare the exact solution to the heuristic solution on a workstation and conclude that our heuristic is very competitive and can be used to reduce the diffusion delay of a broadcast frame in a tree.

The Casemans 2011 workshop opens a platform to researchers of context-aware computing and self-managing systems to investigate the usefulness of context-awareness in emerging applications such as rescue applications, disaster avoidance and overcoming mechanisms, social networking, etc. These applications typically require timely context information to localise people and to share information based on shared interest as well as situations. An interesting research question is how to define and capture mutual context and how to share information in an efficient manner. Hence, the workshop focuses on context acquisition, modelling, reasoning, actuating techniques.

The energy consumption of ICT infrastructures has increased considerably in the recent years. This has resulted in extensive research on dynamic power management strategies as well as data centre design and placement. The main problem with most of the proposed or existing approaches is that they do not fully take the distributed nature of and strong logical dependencies between executed services into account. However, without a comprehensive knowledge of the wider relationships between services, local power management strategies may be ineffectual or can even result in high aggregate energy cost. Understanding this relationship is useful for fine-grained energy-aware computing. For example, services that run on underutilised servers can be stopped or seamlessly migrated to other servers, so that the underutilised servers can be turned off. Alternatively, a re-binding process can be used if the cost of service migration is high. Such advantages can be fully exploited if the dependency between services is properly understood and meaningfully modelled. This paper introduces a conceptual architecture for an energy-aware service execution platform and compares three optimisation mechanisms to support dynamic service migration and rebinding.

This paper investigates the energy cost of control packets in contention-based medium access control protocols in wireless sensor networks. Control packets are useful to avoid collision and overhearing, but cost a significant amount of energy. Therefore, whether or not to apply control packets is a trade-off. It will be shown that this trade-off mainly depends on the packet arrival rate at individual nodes, the transmission rate, the duty cycle and the average number of active neighbors in the network.

This paper provides a qualitative analysis of cross-layer approaches in wireless sensor networks. It begins by presenting two approaches in cross-layer design. The first one focuses on the existence of an auxiliary communication link between the layers - here the OSI layered-approach is being implied - in order to support the exchange of information which is useful for adaptation. The approach itself leaves the layered architecture intact. The second approach takes the OSI layer as a reference, but merges two or more layers together or defines an entirely new dimension to support efficient communication. Therefore, the newly formed architecture requires, in part or wholly, new suit of protocols which may not fit into the layered architecture. The flexibility and performance of the two approaches is compared by investigating some proposed network architectures.

This paper addresses the signal processing aspect of wireless sensor networks. It analyzes several time and frequency domain features of measurements that are taken from 3D accelerometer sensors. The measurements represent various types of movements related to humans and cars. The aim is to obtain quantitative as well as qualitative comparisons concerning the expression power of these features in the presence of various sources of uncertainties (calibration, placement of sensors, and time synchronization). For the qualitative analysis, we define fuzzy sets and fuzzy membership functions for all the features. Particular attention is given to the analysis of the existence of correlation between measurements of different sensor nodes. We will demonstrate that correlation coefficients of both time and frequency domain features exhibit high degrees of uncertainties. On the other hand, short time Fourier transformations (STFT) of all types of movements prove to be agnostic of various forms of measurement and calibration errors.

The growing demand for a wireless link is taking the deployment of wireless local area networks away from the notion of carefully planned and carefully managed setting, into a randomly deployed and independently managed (if at all) network setting. This results in highly contentious networks. In fact research shows that in most metropolitan cities, the size of contentious access points that are closely located with each other are overwhelmingly high. Subsequently, the performance of these networks is highly fluctuating and unpredictable. This paper aims to show how spontaneously deployed networks perform in the presence of fluctuating node densities. We perform an exhaustive simulation for different applications, including, VoIP, HTTP, and FTP. We focus on non-commercial deployments (private use at home or in an apartment).

Context recognition is an essential aspect of intelligent systems and environments. In most cases, the recognition of a context of interest cannot be achieved in a single step. Between measuring a physical phenomenon and the estimation or recognition of what this phenomenon represents, there are several intermediate stages which require a significant computation. Understanding the resource requirements of these steps is vital to determine the feasibility of context recognition on a given device. In this paper, we propose an adaptive context-recognition architecture that accommodates uncertain knowledge to deal with sensed data. The architecture consists of an adaptation component that monitors the capability and workload of a device and dynamically adapts recognition accuracy and processing time. The architecture is implemented for an audio-based context recognition. A detail account of the tradeoff between recognition time and recognition accuracy is provided.

A significant amount of research effort is being carried out by the research community to increase the scope and usefulness of wireless sensor networks; to optimise life time by developing energy efficient power management, self-organising, medium access and routing protocols; and to reduce the cost of sensing nodes so that dense and robust deployment is possible. Though much has already been achieved, currently the cost of commercially available wireless sensor nodes is considerable and the wide applicability of proposed or existing protocols is still under investigation. One essential problem associated with cost or wide applicability of protocols is that sensor networks are application-specific. Protocols and in-network algorithms are optimised for particular sensing tasks. On the other hand, in research environments researchers would like to experiment not with a single application but with many applications. Considering the not-so-cheap sensing nodes available on the market and the management overhead of deploying wireless sensor networks, it is not economical or efficient to dedicate wireless sensor networks just to a single application, not at present at any rate. We therefore propose a middleware that enables researchers to experiment with multiple applications while providing them with essential in-network functionalities to satisfy individual application's requirements. The middleware cleanly separates sensing from network management so that application developers can obtain data from the wireless sensor networks without having to deal with management concerns.

Presents the introductory welcome message from the conference proceedings.

Wireless sensor networks are used for structure monitoring and border surveillance. Typical applications, such as sensors embedded in the outer surface of a pipeline or mounted along the supporting structure of a bridge, feature a linear sensor arrangement. Economical power use of sensor nodes is essential for long-lasting operation. In this paper, we present MERR (minimum energy relay routing), a novel approach to energy-efficient data routing to a single control center in a linear sensor topology. Based on an optimal transmission distance, relay paths are established that aim for minimizing the total power consumption. We study MERR by both stochastic analysis and simulation, comparing it to other possible approaches and a theoretically optimal protocol. We find that MERR consumes 80% less power than conventional approaches and performs close to the theoretical optimum for practicable sensor networks.

This paper presents a distributed architecture for reasoning about a higher-level context as an abstraction of a dynamic real-world situation. Reasoning about a higher-level context entails dealing with data acquired from sensors, which can be inexact, incomplete, and/or uncertain. Inexact sensing arises mostly due to the inherent limitation of sensors to precisely capture a real world phenomenon. Incompleteness is caused by the absence of a mechanism to capture certain real-world aspects; and uncertainty stems from the lack of knowledge about the reliability of the sensing sources, such as their sensing range, accuracy, and resolution. The proposed architecture enables the modeling of a context with facts and beliefs; the model is useful for dealing with data from a variety of sensors with different sensing specifications. It will be shown how the architecture enables the application of empirical knowledge of some physical properties of a place (temperature, relative humidity, sound pressure, light intensity and time) to model and reason about a person's whereabouts. Subsequently, depending on the types and reliability of sensors available at any given time, a mobile device could be able to discriminate between various places-corridors, rooms, buildings, and outdoors-with different degrees of uncertainty