Evaluation of energy consumption and device lifetime of battery-powered wireless sensor networks (WSN) is usually based on measurements or simulations of the total charge (i.e.\ total mA-h) consumed by the device. In reality, batteries are complex electro-physical systems and their discharge behavior depends heavily on the timing and intensity of the applied load. However, there is very little data available about the kinds of batteries and loads that are typically used in WSNs. It is therefore unclear how battery dynamics affect sensor lifetime. We describe a cost-effective testbed for characterizing Li-coin cells and present discharge measurements for over 50 combinations of WSN-typical load parameters. Our results are the first to quantify in-depth the discharge behavior of batteries in the WSN context. We show that loads with very similar time-average currents can differ significantly in the amount of battery capacity that can be used before the battery reaches a cut-off output voltage. In particular, the results suggest that loads with higher current and correspondingly lower duty-cycles and loads with shorter duration are more battery friendly.

Starting from R11, the 3GPP standard includes a complex, network-based offloading policy framework, called Access Network Discovery Selection Function (ANDSF). In our paper, we investigate the effectiveness of ANDSF in geographical areas where LTE cells and WiFi hotspots overlap. We first present a model describing the key properties of multi-RAT networks as visible by the operator and the offloading policy rules that apply to them. Our model captures user behavior and allows us to express any 3GPP policy in a compact and convenient way. We then use the model to develop a dynamic offloading scheme, which is fully compatible with 3GPP specifications and dynamically adapts to changing traffic patterns. We analyse it in a typical two-tier 3GPP scenario, comparing its performance to those of three alternate offloading strategies. We also investigate the effectiveness for data offloading of the current and proposed features of 3GPP ANDSF.

It is well-recognized that our dependence on mobile communications grows; however, users of wireless devices may encounter inadequate coverage due to a variety of shortage and outage circumstances. This is an especially urgent issue in disaster areas where the access to outside world is critical for rescue operations. In this paper, we present a self-organizing communication framework for extending wireless coverage for mobile devices, without requiring modifications of existing wireless infrastructures. Participating devices form a hybrid cellular - mobile ad hoc network, relaying data off net through the nodes that have cellular connections. We design a low-complexity, energy-aware, multi-path data routing mechanism for this framework. The proposed routing includes two major components: a baseline routing that includes a locally reactive and hence low-complexity routing sub-component to handle mobility; and an energy-aware multi-path routing that is motivated by an energy optimization problem and uses only local information. Packet-level simulations show that the proposed routing scheme achieves good performance in relay, packet delivery, and energy consumption. In addition, we implement this framework on Android devices and conduct phone-in-the-loop emulations. The results from the emulation show that the proposed routing scheme can achieve more efficient energy utilization.

The acoustic capabilities (i.e. microphone) and the fast processors of modern smartphones allow for the transmission of data to groups of such devices through the audio channel. We discuss an acoustic data transmission system for broadcast communication to a multitude of smartphones without the need of a radio access point. Acoustic data transmission is particularly attractive in scenarios that involve sound systems (e.g., movie theaters or open-air film festivals). We discuss different techniques to hide data in sound tracks and how to form a microphone array from a collection of smartphones in the same location. Collaborating smartphones share (using their radio interfaces to form an ad hoc network) the received data streams to jointly correct errors. With a testbed of up to four smartphones, we demonstrate how the robustness and reliability of a downlink broadcast via an acoustic communication system can be improved by collaboration between spatially distributed devices. With field tests in different scenarios, we investigate the potential gain of the collaboration in a real environment.

We present a localization system that targets rapid deployment of stationary wireless sensor networks (WSN). The system uses a particle filter to fuse measurements from multiple localization modalities, such as RF ranging, neighbor information or maps, to obtain position estimations with higher accuracy than that of the individual modalities. The system isolates different modalities into separate components which can be included or excluded independently to tailor the system to a specific scenario. We show that position estimations can be improved with our system by combining multiple modalities. We evaluate the performance of the system in both an indoor and outdoor environment using combinations of five different modalities. Using two anchor nodes as reference points and combining all five modalities, we obtain RMS (Root Mean Square) estimation errors of approximately 2.5 m in both cases; while using the components individually results in errors within the range of 3.5 and 9 m.

RESTful services gained a lot of attention recently, even in the enterprise world, which is traditionally more webservice centric. Data centric RESTFul services, as previously mainly known in web environments, established themselves as a second paradigm complementing functional WSDL-based SOA. In the Internet of Things, and in particular when talking about sensor motes, the Constraint Application Protocol (CoAP) is currently in the focus of both research and industry. In the enterprise world a protocol called OData (Open Data Protocol) is becoming the future RESTful data access standard. In this paper we introduce and evaluate an OData implementation for sensor motes and thus allow direct seamless integration of such motes into various enterprise systems. We evaluate the OData protocol in terms of performance and engery consumption, considering different data encodings, and compare it against a pure CoAP implementation. We were able to demonstrate that the additional resources needed for an OData/JSON implementation are reasonable when aiming for enterprise interoperability, where OData is suggested to solve both the semantic and technical interoperability problems we have today when connecting systems.

We study the use of Erasure Codes (ECs) for transmitting information from mobile sensor nodes to stationary base stations. In particular, we are interested in improving the overall communication reliability of the wireless communication. Our scenario is wildlife monitoring in which bats are equipped with tiny sensor nodes, just being capable to store a few kB of data and to exchange information over a wireless communication link. This link is used, on the one hand, for determining contact times between individuals. On the other hand, these contacts are communicated in aggregated form to stationary base stations. Since the channel quality may vary quickly due to the continuous movements of bats and the heterogeneous environment, the communication is in general assumed to be highly unreliable. Conventional reliability improving approaches such as full data replication or on-demand retransmission are too expensive or even not possible due to very strict energy constraints and asymmetric channels. ECs allow to enhance the reliability of data transmissions by transmitting redundant data. In this work, we investigate the trade-off between reliability achieved and the cost in form of additional transmissions, i.e., the additional energy costs. Our results clearly show that ECs improve the communication reliability considerably with almost no impact on the resulting delay.

Evaluation of energy consumption and device lifetime of battery-powered wireless sensor networks (WSN) is usually based on measurements or simulations of the total charge (i.e.\ total mA-h) consumed by the device. In reality, batteries are complex electro-physical systems and their discharge behavior depends heavily on the timing and intensity of the applied load. However, there is very little data available about the kinds of batteries and loads that are typically used in WSNs. It is therefore unclear how battery dynamics affect sensor lifetime. We describe a cost-effective testbed for characterizing Li-coin cells and present discharge measurements for over 50 combinations of WSN-typical load parameters. Our results are the first to quantify in-depth the discharge behavior of batteries in the WSN context. We show that loads with very similar time-average currents can differ significantly in the amount of battery capacity that can be used before the battery reaches a cut-off output voltage. In particular, the results suggest that loads with higher current and correspondingly lower duty-cycles and loads with shorter duration are more battery friendly.

A major challenge in mobile wireless devices for opportunistic networks is to minimize the energy consumption. The minimization however should not come at a cost of reduced application throughput (i.e. goodput). This work evaluates the potential performance gains for mobile nodes that adopt a duty-cycling scheme in an opportunistic context. We present an analytical framework for evaluating the energy consumption of nodes based on a probabilistic estimation of effective contact durations, and we validate this framework on a mobility scenario. We further perform extensive trace-driven simulations and demonstrate that a duty-cycling scheme considerably improves the performance of opportunistic content distribution systems by decreasing the energy consumption without significantly affecting the goodput.

When studying and designing protocols for mobile opportunistic networks, most works consider only direct contact patterns between mobile nodes. Tracking these contacts is important for end-to-end communications but relying only on this information provides a limited view about the transmission capabilities between mobile nodes. Often, mobile users find themselves not in direct contact but at a further distance from one another - referred to as k-contact - that still allows them to communicate. In this work, we study the k-contact opportunities between mobile nodes and try to understand to what extend these k-contacts can be predicted. Using real-world datasets, we provide evidence about the predictable nature of k-contacts and evaluate the benefits of these results compared to direct contact predictions.

Mobile applications are becoming increasingly popular, and with more users connecting to the mobile Web each day, this trend is not going to change anytime soon. As applications and services are starting to target these increasingly mobile users, new challenges arise. In this paper we address the problem of how to effectively evaluate new applications that take into account the geographic location of the users. Experiments with real test subjects are typically expensive and does typically not allow repeatable experiments, needed for fair head-to-head comparisons of alternative protocol implementations. Without a large user population, on which to perform trial runs, system designers are therefore often limited to simulations and mathematical modeling when evaluating these systems. Unfortunately, simulating or modeling realistic scenario for mobile networks is complicated and time consuming. In this paper we present the design of a simple emulation framework for performance evaluation and testing of mobile applications in realistic scenarios. Our simple in-house emulation testbed combines production hardware and software to allow emulation of realistic and repeatable mobility scenarios, in which the mobile user can travel long distances, while being served by an application server. The framework allows (i) geo-location information, (ii) client network conditions such as bandwidth and loss rate, as well as (iii) the application workload to be emulated synchronously. To illustrate the power of the framework we also present the design, proof-of-concept implementation, and evaluation of a geo-smart scheduler for application updates in smartphones. This geo-smart scheduler reduces the average download time by using a network performance map to schedule the downloads when at places with relatively good conditions. Our trace-driven evaluation of the geo-smart scheduler, illustrates the workings of the emulation framework, and the potential of the geo-smart scheduler.

Practical intrusion detection in Wireless Multihop Networks (WMNs) is a hard challenge. By design, base stations---or nodes---in a WMN aim at being self organized, and need little manual intervention. This self-organized and distributed nature of WMNs makes centralized intrusion detection difficult. On top of that, resource constraints of the nodes and the characteristics of the wireless medium often render node-based approaches impractical. It has been shown that an active-probing-based network intrusion detection system (AP-NIDS) is practical for WMNs. However, understanding its interworking with real networks is still an unexplored challenge. In this paper, we investigate this challenge in practice. We identify the general functional parameters that can be controlled in the active-probing technique, for example, number of repetitions of testing packets, repetitions of control packets, and time intervals between transmissions. By means of extensive experimentation, we tune these parameters and analyze the trade-offs between them, aiming at reducing false positives, overhead, and detection time. We study their behavior with and without a congested network. The traces we collected helps us to understand when and why the active probing fails, and let us present countermeasures to prevent it.

An Accident Warning System (AWS) is a safety application that provides collision avoidance notifications for next generation vehicles with a Vehicular Ad-hoc Network (VANET) providing the communication functionality to exchange these notifications. Even though researchers have been investigating how to design better AWSs for more than half a decade there are still plenty of grey areas and something of a lack of substantial requirements analysis. In order to build a practical warning system, it is important to ascertain the system requirements and necessary cooperation characteristics between complementary devices. This paper presents a practical model of an accident warning system by stipulating the requirements in a realistic manner and thoroughly reviewing existing proposals with a view to identify gaps in this area.

Floating Car Data (FCD) consist of information generated by moving vehicles and uploaded to Internet-based control centers for processing and analysis. As upcoming mobile services based on or built for networked vehicles largely rely on uplink transfers of small-sized but high-frequency messages, FCD traffic is expected to become increasingly common in the next few years. Presently, FCD are managed through a traditional cellular network paradigm: however, the scalability of such a model is unclear in the face of massive FCD upload, involving large fractions of the vehicles over short time intervals. In this paper, we explore the use of vehicle-to-vehicle (V2V) communication to partially relieve the cellular infrastructure from FCD traffic. Specifically, we study the performance boundaries of FCD offloading in presence of best- and worst-case data aggregation scenarios, as achieved by optimal and practical solutions.

Inter-Vehicle Communication is expected to be widely adopted during the next years by the car industry, enabling Vehicle-to-Vehicle and Vehicle-to-Infrastructure communication targeting at safer commuting. Therefore, the research community have been working towards providing a robust communication solution that will interoperate with existing network infrastructure and will provide an effective way of communicating using location information. One of the proposed solutions is the geonetworking protocol, standardized by ETSI, using location based addresses and an IPv6 adaptation sublayer for communicating with the Internet. Several implementations of the protocol exist currently, however most of them are integrated in simulation environments, thus ignoring several factors induced when experimenting under real conditions. In this paper we present our own implementation of an ETSI compliant geonetworking protocol and its integration in the NITOS Future Internet facility that enables open and remote access to experimenters on a 24/7 basis. This is the first real implementation in an open wireless testbed, which offers to the research community high diversity in the configuration parameters of their ITS experiments. We further explain our extensions that enable real world testing of our protocol implementation and we finally evaluate our solution in a real large scale wireless setup.

Vehicles are getting increasingly connected. Several technologies have emerged over the last decade that allow cars to communicate with each other and with the Internet. In this paper we propose to use the new Bluetooth Low Energy (BLE) standard as an alternative technology to exchange data between vehicles. By the means of experiments we show that smartphones with BLE radios can be used to send information at low latency from one car to another even while driving. A communication range of up to 100m can be achieved depending on the scenario and environment. Those promising first results are then used as basis for discussion to identify the potential of BLE for different types of vehicular applications.

Simulation of cellular network communication is complex and typically requires a high degree of knowledge about the underlying network and its parameters. At the same time simulating cellular networks is important for the automotive industry in order to be able to test the feasibility of applications that use car-to-x-communication applications before performing costly field tests. In this paper we propose a trace-based simulation model derived from real-world measurements. It does not require any information about the network besides information that can be readily measured by a regular user, it is much faster than regular simulation, and it has been validated by comparing simulation results to real world measurements.

Quickly identifying the severity of highway acci- dents, as well as the resources required to assist the people involved in those accidents, is a basic requirement for future intelligent transportation systems. In this context, vehicular communication technologies currently being standardized are able to provide novel solutions to address this problem. In this work we study the feasibility of combining vehicle- to-vehicle (V2V) and vehicle-to-infrastructure (V2I) communica- tions to deliver a video stream from the place of the accident to the traffic authorities. Our approach relies on vehicles as data relays, thus having the additional advantage of providing drivers with a clear view about the accident, thereby helping to reduce stress and improving traffic flow. An experimental analysis comparing different traffic flooding mechanisms for wireless networks show that the proposed system is viable for highways with moderate/high amounts of traffic, although highlighting the need for more efficient mechanisms specifically addressing broadcast propagation in highway envi- ronments.

Road traffic monitoring is one of the key applica- tions in the Intelligent Transport System field. New technologies are now provided in this field and among the most relevant ones there is the DSRC (Dedicated Short Range Communication) set of protocols and standards where vehicles wirelessly communicate. In this paper, we deal with the application of Vehicular Ad-Hoc Networks to road traffic monitoring and we present the design of two distributed protocols based on the DSRC. A realistic sim- ulation of a main expressway in Rome, Italy, is implemented and the performances of the two proposed monitoring methodologies are evaluated in case of regular traffic conditions and in case of a car accident. In both cases the protocols are able to capture in a very quick time (few seconds) the current traffic conditions even on a quite long road of about 70 km. A discussion about the impact of the market penetration rate of the on-board DSRC devices on the protocols performance is also provided.

Driving vehicles in platoons has the potential to improve traffic efficiency, increase safety, reduce fuel consumption, and make driving experience more enjoyable. A lot of effort is being spent in the development of technologies, like radars, enabling automated cruise control following and ensuring emergency braking if the driver does not react in time; but these technologies alone do not empower real platooning. The initial idea of building dedicated infrastructures for platoons, has been set aside favoring the philosophy that foresees scenarios, where automated vehicles share the road with human-driven ones. This arises interesting new questions regarding the interactions between the two categories of vehicles. In this paper we focus on the analysis of interferences caused by non-automated vehicles during a Join maneuver. We define the application layer protocol to support the maneuver, together with situations that can prevent successful termination, and describe how they can be detected. The validity of the approach is proven by means of simulations, showing either that the maneuver can successfully be performed, or safely be aborted. Finally, we analyze the impact of the Packet Error Rate on the failure rate of the maneuver, showing that packet losses mainly affect the maneuver from a coordination point of view, rather than stability of the system, i.e., even at high loss rates, cars never violated a minimum safety distance.

It has been shown that paragliders could considerably benefit from ad-hoc communication, be it for safety or the prolonging of flight times through the exchange of thermal information. The simulation of these so called Flying Ad-Hoc Networks (FANETs) can help evaluate the feasibility and performance of these and other applications inexpensively and at large scale. Their specific communication characteristics, caused by inevitable suboptimal antenna placement and 3D node distribution, require accurate channel and propagation models in order to produce meaningful results in a simulation environment. We identify two important parameters that heavily influence the Received Signal Strength (RSS), namely the vertical angle between paragliders and their horizontal relative bearing. Based on extensive real life experiments we present a deterministic and computationally inexpensive radio propagation model that is able to reliably predict our measurements. Our work allows the realistic simulation of wireless communication between paragliders.

Given the broad acceptance of the DSRC/WAVE protocol stack in the vehicular networking community, both the automotive industry and the scientific community are working towards so-called "day one" applications. Currently, large scale field operational tests are going on to assess the performance of developed protocols and applications. Still, the key technique for performance evaluation is simulation. Accurate microscopic simulation of Inter-Vehicle Communication (IVC) is needed, especially for safety critical applications. This is reflected in many recent publications trying to push this in terms of radio shadowing models, signal propagation, etc. Still, it is not fully understood to characterize some effects given the constraints in terms of simulation time and performance. We concentrate on the fading model. Simulating freeway scenarios, the two ray interference model is considered the base line, but what about suburban and city scenarios? This paper looks into this investigating, for the first time, the impact of the street width, i.e., distance between buildings, and its relation to the correct use of propagation models. We conducted different measurement campaigns on streets with different widths and compare the results to theoretic models that are frequently used for IVC studies. The most prominent result is that we discovered a clear difference when assessing safety applications in wide streets compared to the narrow streets.

Media streaming in automotive environments is becoming more important with the proliferation of 3G/4G technologies and the general demand for consuming internet content in cars. Especially the rising popularity of Music on Demand and Media Cloud Storage services pushes automotive manufactures efforts to provide decent music streaming capabilities in vehicles. This fact has recently brought car manufactures and music streaming services together. Thanks to today's mobile broad band Internet connectivity, music streaming is becoming available in the car. Volvo and Ford have announced to pair up with the popular music streaming service Spotify. Ford does already have a partnership with Rhapsody's music streaming and with the cloud music service Amazon Cloud Player while BMW is going to bring Rara to their vehicles. With the increasing usage of those services in cars in the future, the service provider will face network problems and so users will experience more and more usability drawbacks when listening to their music. These issues on the one hand arise from the fact, that the mobile network load will be easily exhausted in crowded areas with many users listening music in their cars. On the other hand, cars frequently need to pass spots of bad reception like tunnels, underpasses or street canyons in urban areas. Whereas in rural areas totally disconnected regions and frequent handovers due to higher driving speed need to be taken into account. In order to tackle such issues, new prefetching algorithms are needed that can bridge connectivity interruptions and prevent network overload. Cars are not as personalized end devices as smartphones, since various persons can be the driver, or the user respectively. This is one reason why simple pre-buffering of personalized song lists does not help here, as it might waste a lot of bandwidth and local storage when downloading lots of songs which will possibly not be heard. Consequently, suitable prefetching algorithms need to consider network coverage along driving route, as well as current network usage information. In this way, cache utilization and bandwidth consumption can be reduced to a minimum. Moreover, the load can be balanced for all users, throughout the available mobile network. The development and assessment of the named class of smart prefetching algorithms is quite complex as they involve a variety of external information and require greater coordination efforts between vehicles. For this purpose, we present a comprehensive simulation environment, which provides the relevant information input and simulates interaction of a greater number of vehicles with a cellular network. Our existing simulation framework VSimRTI couples discrete event simulators from different fields. VSimRTI is already capable to simulate vehicular traffic, application logic and communication between the vehicles via ad-hoc networks by coupling the relevant simulators. In this work, we need to extend this tool for the feasibility to simulate cellular communication between vehicles and the according multi-media server. Therefore, a new cellular communication simulator has been developed and coupled to VSimRTI. This simulator VSimRTI\_Cell especially focusses on simulating the transmission delay through the wireless and wired parts of the network. The major advantage of our approach is the independence of the wireless access technologies (3G, 4G). Moreover, the setup can respect for external cellular users (e.g. in crowded areas) which are not equipped in our simulation, but still consume a certain amount of the shared cell bandwidth. Even though, the implemented communication models abstract from different details, they are adequate for evaluations from application perspective. The main contribution, presented in this paper, is our simulation environment and evaluation methodology for smart prefetching algorithms, not the algorithms themselves. Therefore, we employ algorithms which are very basic, but involve all necessary aspects of smart prefetching. For future work, we plan on advancing our algorithms by including communication of meta data among the vehicles for load balancing and user side content access prediction like title skipping in playlists. Finally, we will show the results of an exemplary evaluation study which compares different basic algorithms for prefetching music content. By these results, we will show that our simulation environment is capable of mapping the relation of bandwidth utilization depending on the number of vehicles downloading media and its impact on user experience. We will further present an evaluation methodology for prefetching algorithms, which is able expose the benefits of such algorithms.

Community Wireless Networks can be a way to make "Internet access for everyone" possible, by sharing a broadband Internet connection via WLAN. The underlying idea is to freely provide network access to anybody in the vicinity of the wireless access point via a Lower-than-Best-Effort (LBE) service, such that non-paying users interfere as little as possible with the "regular" Internet usage. Such a service faces challenges at various network layers; this paper discusses some of them, focusing on layers 2, 3 and 4.

Wireless Community Networks in the developing world satisfy the basic needs of remote users to information access. However, community networks in developing regions usually rely on low-bandwidth backhaul links that are shared amongst a large user base, driving these links to sub-packet regimes where the per-flow throughput is less than one packet per RTT. TCP performance significantly degrades in such conditions, resulting in severe unfairness and high packet loss rates. In this paper, we investigate the performance of scavenger transport methods, namely LEDBAT and its fair modification fLEDBAT, in the sub-packet regime of shared backhaul links in developing regions. Our intention is to explore the feasibility of using such scavenger transport methods for uploading content over bandwidth constrained backhauls. Our findings show that LEDBAT achieves higher link efficiency and fairness compared to TCP in a variety of sub-packet regime scenarios. When TCP and LEDBAT flows share the same link in the sub-packet regime, LEDBAT flows are more aggressive, consuming more resources than TCP. Therefore, we conclude that a more conservative strategy after consecutive timeouts and shared bottleneck detection mechanisms need to be incorporated into the core LEDBAT algorithm, in order to correctly adjust its congestion window in the sub-packet regime.

We present the Liberouter framework, a complete communication system design (and its implementation) to enable neighborhood networking without relying on the Internet infrastructure for information exchange. The core is a low-cost router platform that serves as WLAN access points, individual or (dis)connected ones, and offers message storage and relaying in combination with a distributed app store. It allows bootstrapping Android devices to become additional routers (and thus expand the network) and install applications. It also instruments other mobile devices to assist in message forwarding and offers web-based access to content of the neighborhood stored locally.