This paper presents conditional probabilistic modeling suitable to characterize the temporal variation of links in wireless body area networks (BAN); according to short, medium and long term fading characteristics. The approach captures first and second order statistics appropriately; and using conditional probabilities, predicts what signal power levels can be expected from 10ms ahead to many seconds into the future in the context of typical channel coherence times. Hundreds of hours of link measurements, in "Everyday" mixed activity BAN, are used to generate these conditional models. We show that such modeling has an important effect to higher level simulation results (e.g., packets received at the application layer) for different simulation scenarios. Moreover we show that the model is computationally efficient as its introduction adds only 8% of simulation time on average. It is also shown that short and medium-term conditional modeling can vary considerably from long-term modeling, particularly given lower instantaneous path loss.

We investigate the impact of wireless channel temporal variations on the design of medium access control (MAC) protocols for body area networks (BANs). Our measurements-based channel model captures large and small time-scale signal correlations, giving an accurate picture of the signal variation, specifically, the deep fades which are the features that mostly affect the behavior of the MAC. We test the effect of the channel model on the performance of the 802.15.4 MAC both in contention access mode and TDMA access mode. We show that there are considerable differences in the performance of the MAC compared to simulations that do not model channel temporal variation. Furthermore, explaining the behavior of the MAC under a temporal varying channel, we can suggest specific design choices for the emerging BAN MAC standard.

We investigate the potential of wireless sensor networks to monitor the structural health of bridges. Needing to explore different sensing, networking, and distributed computing approaches in real conditions, we are led to the creation of an on-site testbed. In this paper we describe the design and implementation of the testbed by offering an overview of the architectural choices and components, discussing obstacles found, and describing the software used to manage it. Finally, we discuss our first experiences with the testbed and the first validation tests of the sensing data gathered.

We consider a body area network (BAN) setting in which sensor nodes send data to a common hub regularly on a TDMA basis, as defined by the emerging IEEE 802.15.6 BAN standard. Our previous work has established the benefits of variable TDMA scheduling in this setting, where the order of communication with nodes in each TDMA round is determined on the fly by the hub, based on the actual outcomes (success or failure) of their recent transmissions. This technique takes into account the very slow fading nature of wireless channels in BANs, and can minimize the expected rate of transmission losses merely by ordering nodes according to their most recent channel state information, without any additional energy overhead. In this paper, we focus on an extended TDMA setting where the length of each TDMA round is greater than the number of nodes, i.e. a number of 'spare' slots are reserved for retransmissions of lost packets. We consider delay-constrained data transmissions based on a Markov model of the wireless channels. We evaluate several variable scheduling strategies in terms of the trade-off between the energy consumption due to the additional retransmissions and the reduction of the loss rate. We show how the best strategy depends on the energy budget of the sensor nodes, and demonstrate the considerable reduction in the loss rate (up to 18%) that can be achieved over a naive approach where static TDMA scheduling is combined with retransmission slots at the end of the round.

We investigate the effect of radio modeling on the simulation results of low-power wireless networks. In particular we focus on the modeling of carrier sensing and state transition delays and describe how these aspects of low-power radios are captured in the Castalia simulator. These aspects are usually neglected by network-level simulators yet they can have a significant effect on the higher level simulation results. We show these effects by exploring a simple simulation scenario: a clique of N nodes -all in range with each other- with N-1 transmitter nodes sending packets to one receiver node, employing the simple CSMA/CA MAC protocol. Varying the sending rate and number of nodes, we show that the throughput at the receiver node can greatly vary depending on the assumptions we make about the radio. We also show that simple decisions at the MAC level, such as the back-off duration and the method of transmitting buffered packets, have a much more pronounced effect on performance when the radio is modeled more accurately.

We consider a typical body area network (BAN) setting in which sensor nodes send data to a common hub regularly on a TDMA basis, as defined by the emerging IEEE 802.15.6 BAN standard. We explore variable TDMA scheduling techniques that allow the order of transmissions within each TDMA round to be decided on the fly, rather than be fixed in advance. Our approach is related to opportunistic scheduling used in other multiuser wireless systems, which aims to maximize the system throughput by allocating transmission opportunities to users with the `best' instantaneous channel. However, the energy overheads of opportunistic scheduling, which requires the continuous reporting of channel states to the scheduler, make it incompatible with body area networks where nodes turn their radio off (`sleep') until it is their turn to transmit, as an energy-saving measure. Accordingly, we consider variable TDMA scheduling strategies that decide on the transmission order at the start of each TDMA round based on the last known channel state of each sensor. We formulate the optimization problem of maximizing the expected throughput (or, equivalently, minimizing the expected loss rate) based on a Markov model of the wireless channel, and present some simple approximate heuristics that can be computed quickly yet perform well in practice. We evaluate the performance of these strategies and heuristics numerically in a wide range of scenarios, and demonstrate the marked improvement they achieve over a fixed TDMA allocation.

We describe our experience from the implementation of the T-MAC protocol for wireless sensor networks in the open-source Castalia simulator. Notwithstanding the popularity of the protocol in the research literature in recent years, we find several practical issues that are not addressed in the original protocol description, which lead to a degree of freedom in the protocol design and implementation and have an impact on its resulting performance. These issues include the ability of the underlying physical layer and hardware to efficiently detect the activation events in the protocol, and necessary changes to the collision resolution and clock synchronization procedures in the presence of varying sleep patterns. Our results highlight the need for rigorous detail in protocol descriptions in the research literature and provide important insights into some of the common pitfalls.

We present Castalia, a simulator for WSN that models many aspects of the WSN system and uses advanced models especially in terms of the channel and radio behaviour. We show the effects of these features in distributed algorithms that work fine with simpler simulators but fail under Castalia. The demo will present the differences, explain the failures and show how to redesign the algorithms to make them work under more realistic conditions.

The paper presents our efforts to validate some high-level aspects of the WSN simulator we have built as well as the operational functionality of our multi-parameter MAC protocol. In order to do so, we resort to real deployments involving TelosB motes. The simulator, named Castalia, boasts the most accurate wireless channel and radio models for WSN found in current literature. These models are capturing some essential experimental findings. This does not guaranty though that the simulator will behave similarly with a real deployment at the high level (i.e., the protocol or application level). We investigate how our multi-parameter MAC protocol behaves in a real deployment so as to take a first step towards validating and possibly tuning Castalia. The investigation starts by determining the connectivity map for the real deployment and then trying to reproduce it in the simulator. We then proceed with the protocol testing and comparing. We report the difficulties faced and our findings from this process.

Sensor Networks need the ability to be dynamically programmed efficiently. Several frameworks have been developed for this purpose largely falling into one of three classes: i) user-guided code updates, ii) database model, and iii) agent-based model. Even though the agent-based model is being touted to bring increased efficiency in terms of energy consumed, its advantage has not been clearly articulated by the relevant sensor network literature. We identify a suitable application to showcase the advantages of autonomously mobile code and analyze the generic solutions offered by the three different classes of frameworks to show that indeed an agent-based approach can bring considerable energy savings for some application scenarios.

Conserving energy has been known as the most significant problem in all facets of sensor network operation. Particularly in routing, researchers were concerned with the problems of using topology control to achieve power efficient routes, as well as finding alternate routes to extend the lifetime of the network. Current schemes though, fail in one or more of the following areas: 1) provide solutions that treat both the aforementioned problems, 2) provide distributed algorithms, 3) account for the overhead of the algorithms 4) compare the results with theoretically computed optimums, 5) account for the effect of the MAC layer. We propose and evaluate a unicast routing algorithm that exploits the ability of the nodes to transmit at multiple power levels. It can find the optimal power-efficient route between two nodes, with less energy and time overheads than the distributed Bellman-Ford algorithm, as well as use alternate routes to extend the total lifetime of the network, up to 87% of the theoretical optimum lifetime, taking into account all overheads. Furthermore, we address many practical considerations in the context of sensor networks. We built our case around three main pillars: i) a distributed algorithm, ii) extensive evaluation of the overhead of the algorithm, and iii) account for the effect of the MAC layer.

Energy related research in wireless ad hoc sensor networks (WASNs) is focusing on energy saving techniques in the application-, protocol-, service-, or hardware-level. Little has been done to manage the finite amount of energy for a given (possibly optimally-designed) set of applications, protocols and hardware. Given multiple candidate applications (i.e., distributed algorithms in a WASN) of different energy costs and different user rewards, how does one manage a finite energy amount? Where does one provide energy, so as to maximize the useful work done (i.e., maximize user rewards)? We formulate the problem at the node-level, by having system-level “hints” from the applications. In order to tackle the central problem we first identify the energy consumption patterns of applications in WASNs, we propose ways for real-time measurements of the energy consumption by individual applications, and we solve the problem of estimating the extra energy consumption that a new application brings to a set of executing applications. Having these tools at our disposal, and by properly abstracting the problem we present an optimal admission control policy and a post-admission policing mechanism at the node-level. The admission policy can achieve up to 48% increase in user rewards compared to the absence of energy management, for a variety of application mixes.

Wireless ad hoc sensor networks have emerged as one of the key growth areas for wireless networking and computing technologies. So far these networks/systems have been designed with static and custom architectures for specific tasks, thus providing inflexible operation and interaction capabilities. Our vision is to create sensor networks that are open to multiple transient users with dynamic needs. Working towards this vision, we propose a framework to define and support lightweight and mobile control scripts that allow the computation, communication, and sensing resources at the sensor nodes to be efficiently harnessed in an application-specific fashion. The replication/migration of such scripts in several sensor nodes allows the dynamic deployment of distributed algorithms into the network. Our framework, SensorWare, defines, creates, dynamically deploys, and supports such scripts. Our implementation of SensorWare occupies less than 180Kbytes of code memory and thus easily fits into several sensor node platforms. Extensive delay measurements on our iPAQ-based prototype sensor node platform reveal the small overhead of SensorWare to the algorithms (less than 0.3msec in most high-level operations). In return the programmer of the sensor network receives compactness of code, abstraction services for all of the node's modules, and in-built multi-user support. SensorWare with its features apart from making dynamic programming possible it also makes it easy and efficient without restricting the expressiveness of the algorithms.

Providing multiple modes to support dynamically changing environments, standards, and new services is prevalent in embedded systems, especially in mobile radio systems. Because such a system frequently contains time-constrained tasks, it is important to analyze the temporal requirements as well as the functional correctness. This paper presents a method to analyze temporal requirements imposed on an embedded real-time system supporting multiple modes. While most performance analysis methods focus only on testing the feasibility of a task or a system, our method goes further by addressing the problem of locating hot spots of a system thereby helping the designer to choose among alternative designs or architectures. We formally define the analysis problem and show that it is very unlikely to be solved efficiently. We present a heuristic algorithm, which is accurate and fast enough to be used in iterative processes in system-level analysis and design. The analysis problem is extended to accommodate probabilistic behavior exhibited by soft real-time tasks.

The quality of wireless links suffers from temporally and spatially varying channel degradations such as noise, interference, and multipath fading. A proper addressing of these impairments is crucial to sustaining quality of service (QoS) in wireless integrated service packet networks such as wireless ATM. Newer radios that allow parameters such as processing gain, symbol rate, and transmit power to be varied on a per packet basis offer interesting opportunities. This paper explores a highly adaptive approach to wireless link control where the various physical and link layer parameters are continually adapted in response to channel and traffic variations. Furthermore, we also explore adaptivity at the upper layer of the protocol stack, since applications typically can not cope with residual distortions in the packet flow. It is illustrated that lower and upper layer adaptation strategies are tightly intertwined, such that they should be managed jointly. Node and basestation architectures to efficiently support such adaptivity are also described.