Giacomo Ghidini

An Energy-Efficient Markov Chain-Based Randomized Duty Cycling Scheme for Wireless Sensor Networks

To extend the life time of a wireless sensor network, sensor nodes usually switch between dormant and active states according to a duty cycling scheme. In randomized schemes, sensors use only partial or no information about their neighbors, and rely on randomness to generate working schedules. Duty cycling schemes are often evaluated in terms of the connection delay, i.e., the time until two neighboring nodes are simultaneously active, and the connection duration, i.e., the time until at least one of them switches to the dormant state. In this paper, we argue that duty cycling time (energy) efficiency, i.e., the ratio of time (energy) employed in ancillary operations when switching from and into deep sleep mode, is an important performance metric too. We present experimental results using Sun SPOT sensors that support our claim and highlight the performance trade-off between connection delay and time (energy) efficiency for a traditional scheme based on independent and identically distributed (i.i.d.) random variables. We propose a novel randomized duty cycling scheme based on Markov chains with the goal of (i) reducing the connection delay, while maintaining a given time (energy) efficiency, or (ii) keeping a constant connection delay, while increasing the time (energy) efficiency. The proposed scheme is analyzed mathematically by deriving the time efficiency, connection delay and duration in terms of the time slot length, duty cycle, and cost of set up and tear down operations. Analytical results demonstrate that the Markov chain-based scheme can improve the performance in terms of connection delay without affecting the time efficiency, or vice versa, as opposed to the trade-off observed in traditional schemes. Experimental results using Sun SPOT sensor nodes with the minimum number of operations during transitions from and into deep sleep mode confirm the mathematical analysis of the proposed Markov chain-based randomized scheme.

A novel framework for energy-efficient data gathering with random coverage in wireless sensor networks

In wireless sensor networks, different applications feature different requirements in terms of such performance metrics as sensing coverage and data reporting latency. In most applications, it is usually sufficient to provide a Desired Sensing Coverage (DSC) lower than full coverage at any instance with the guarantee that the whole area will eventually be covered within a specified delay bound. Due to the fact that these applications are also expected to run for longer periods of time and at the same time battery recharging and replacement are costly, energy consumption in wireless sensor networks should be minimized while achieving the application goals. In this article, we propose a novel framework for application-specific data gathering which exploits a trade-off between coverage and latency, thereby minimizing energy consumption and extending the network lifetime. The proposed energy-efficient, constant-time, randomized scheme, called Coverage-Adaptive raNdom SEnsor sElection (CANSEE), selects a subset of k sensors to report at each round so as to fulfill the application-specific requirement of desired sensing coverage and bounded latency, instead of always guaranteeing full coverage and minimum latency. We present a probabilistic model to estimate: (i) the connectivity of those selected k sensors and the number of additional sensors needed to guarantee connectivity; (ii) a lower bound on k in each round; and (iii) the probability of almost surely having k data reporters using the Chernoff bound. The immediate event detection capability achieved by the proposed CANSEE scheme is also analyzed to compare the performance of our framework with other data gathering schemes that allow 100% coverage. Simulation results demonstrate that our framework leads to a significant conservation of energy (and thus extended network lifetime) with a small trade-off between coverage and data reporting latency, yet providing the required data reporting capability.

Fuseviz: A framework for web-based data fusion and visualization in smart environments

Recent advances in technology and algorithms for smart environments have made it possible to collect and store large amounts of data about many aspects of human life and the surrounding environment with limited effort and cost. However, such data become useful to lay users with no background in data analysis only if they are presented in a fashion that supports intuitive interaction to spot the patterns and trends, thus transforming the data into valuable information. In this paper, we introduce FuseViz, a framework for Web-based fusion and visualization of data in smart environments. FuseViz addresses the challenges posed by large, live, heterogeneous, and dynamic data streams from autonomous data sources, and lay users, with two basic features: fusion and visualization. CouchDB, a schemaless database with a ReSTful API and MapReduce support, is used to fuse data streams from multiple sources, while Web-based visualization is implemented on top of D3, a JavaScript library for manipulation of data-driven documents. We demonstrate the capabilities of FuseViz with E2Home, a case study application for energy-efficient smart home environments. We show how the precise information provided by E2Home can help the user easily improve the home energy efficiency by more than 10%.

Localization and scheduling protocols for actor-centric sensor networks

We propose novel localization and routing protocols in an actor-centric wireless sensor network consisting of an actor node and a large number of energy-constrained sensors operating under L different periodic sleep–awake schedules. Specifically, we propose a semidistributed localization algorithm in which a small subset of sensors extracts their positions in polar coordinates based on the messages received from the actor, and subsequently localizes (also in polar coordinates) the remaining sensors. By modeling the deployed sensors as a two-dimensional Poisson point process and applying well-known results from the coupon collectors problem and Chernoff bounds, we analytically derive and also validate, by simulation, the sensor density required to localize all sensors in the network with high probability. The actor-centric network can be modeled by a cluster adjacency graph G with the help of the already localized polar coordinates that logically partition the network into concentric coronas (around the actor), each subdivided in a varying number of clusters (of almost the same area). To avoid intercluster collisions in G, sensors in different clusters transmit on different channels. A lower bound on the number of channels required to schedule the transmissions without collisions is obtained by solving a distance-2 vertex coloring problem on G. Optimal and quasioptimal fully distributed algorithms are provided to determine the channel assigned to each cluster in constant time. Finally, we apply these results to develop a geographic routing protocol: the messages generated from the sensors in a given cluster are routed toward the actor through the unique shortest path of G that starts from the node associated with the cluster and goes up to the corona where the actor resides. In each cluster, to avoid redundant retransmissions toward the actor, we select L leaders, one for each periodic sleep–awake schedule.

Energy-efficient markov chain-based duty cycling schemes for greener wireless sensor networks

To extend the lifetime of a wireless sensor network, sensor nodes usually duty cycle between dormant and active states. Duty cycling schemes are often evaluated in terms of connection delay, connection duration, and duty cycle. In this article, we show with experiments on Sun SPOT sensors that duty cycling time (energy) efficiency, that is, the ratio of time (energy) employed in ancillary operations when switching from and into deep sleep mode, is an important performance metric too. We propose a novel randomized duty cycling scheme based on Markov chains with the goal of (i) reducing the connection delay, while maintaining a given time (energy) efficiency, or (ii) keeping a constant connection delay, while increasing the time (energy) efficiency. Analytical and experimental results demonstrate that the Markov chain-based scheme can improve the performance in terms of connection delay without affecting the time efficiency, or vice versa, as opposed to the trade-off observed in traditional schemes. We extend the proposed duty cycling scheme to a partially randomized scheme, where wireless nodes can switch into active state beyond their schedules when their neighbors are active to anticipate message forwarding. The analytical and experimental results confirm the relationship between connection delay and time efficiency also for this scheme.

A semi-distributed localization protocol for wireless sensor and actor networks

We consider a wireless sensor and actor network (WSAN) consisting of a large number of tiny, low-cost sensors uniformly and independently distributed in a two-dimensional geographical region around a few powerful entities, called actors. To save energy, the sensors operate according to sleep/awake schedules in an asynchronous manner. In this setting, we propose a semi-distributed, actor-centric localization algorithm which organizes the sensors in the vicinity of each actor by means of a discrete polar coordinate system. Specifically, each sensor is localized when it acquires the corona and sector coordinates of the region it resides in. To accomplish the localization task, each actor first trains a subset of sensors in its vicinity, which in turn train their neighbors. By modeling the deployed sensors as a two-dimensional Poisson point process and applying well-known results from the Coupon Collectors problem and Chernoff bounds, we derive bounds on the sensor density required to localize with high probability all sensors in the actors vicinity. Finally, we verify the analytical bounds with results from our simulation experiments.

Online stream processing of machine-to-machine communications traffic: A platform comparison

In a machine-to-machine (M2M) communications system, the deployed devices relay data from on-board sensors to a back-end application over a wireless network. Since the cellular network provides very good coverage (especially in inhabited areas) and is relatively inexpensive, commercial M2M applications often prefer it to other technologies such as WiFi or satellite links. Unfortunately, having been originally designed with human users in mind, the cellular network provides little support to monitor millions of unattended devices. For this reason, it is extremely important to monitor the underlying signalling traffic to detect misbehaving devices or network problems. In the cellular network used by M2M communications systems, the network elements communicate using the Signalling System #7 (SS7), and a real-life system can generate tens of millions of SS7 messages per hour. This paper reports the results of our practical investigation on the possibility to use distributed stream processing systems (DSPSs) to perform real-time analysis of SS7 traffic in a commercial M2M communications system consisting of hundreds of thousands of devices. Through a thorough experimental evaluation based on the analysis of real-world SS7 traces, we present and compare the implementations of a DSPS-based data analysis application on top of either the well-known Storm DSPS or the Quasit middleware. The results show that, by using DSPS services, we are able to largely meet the real-time processing requirements of our use-case scenario.

Soft real-time GPRS traffic analytics for commercial M2M communications using spark

Commercial applications of wireless sensor networks, also known as machine-to-machine (M2M) communications, feature hundreds of thousands or even millions of devices. These M2M applications often rely on cellular networks like GSM that were not designed with such use cases in mind. Based on our first-hand experience at a large provider of M2M communications solutions, there is a need for soft real-time traffic analytics solutions to help engineers monitor and manage the millions of devices deployed in these M2M applications. We present a solution for soft real-time GPRS traffic analytics built on Apache Spark, a framework for distributed in-memory computing. The proposed solution captures GPRS traffic, processes it, and decorates it with details about the devices, networks, and M2M applications. It then computes a whole array of statistics that are presented in charts and maps on a live Web application dashboard, or may be fed to other systems for data mining. In a series of experiments, previously captured GPRS traffic from real-life commercial M2M applications is played back to the traffic analytics solution at different rates, and is processed on clusters of varying size. Results show that our solution handles GPRS traffic rates of 3,333 packets/sec, which are 2X the rates of an M2M application with close to one million devices, with a latency below one minute on a Spark cluster with four m1.large slave instances in Amazon EC2 at a cost of

Advancing M2M communications management: A cloud-based system for cellular traffic analysis

7,665/year. These costs can be reduced to approx.

Broadcast analysis in dense duty-cycle sensor networks

700/year by bidding on SPOT instances.