Angelo Brayner

An adaptive in-network aggregation operator for query processing in wireless sensor networks

A wireless sensor network (WSN) is composed of tens or hundreds of spatially distributed autonomous nodes, called sensors. Sensors are devices used to collect data from the environment related to the detection or measurement of physical phenomena. In fact, a WSN consists of groups of sensors where each group is responsible for providing information about one or more physical phenomena (e.g., group for collecting temperature data). Sensors are limited in power, computational capacity, and memory. Therefore, a query engine and query operators for processing queries in WSNs should be able to handle resource limitations such as memory and battery life. Adaptability has been explored as an alternative approach when dealing with these conditions. Adaptive query operators (algorithms) can adjust their behavior in response to specific events that take place during data processing. In this paper, we propose an adaptive in-network aggregation operator for query processing in sensor nodes of a WSN, called ADAGA (ADaptive AGgregation Algorithm for sensor networks). The ADAGA adapts its behavior according to memory and energy usage by dynamically adjusting data-collection and data-sending time intervals. ADAGA can correctly aggregate data in WSNs with packet replication. Moreover, ADAGA is able to predict non-performed detection values by analyzing collected values. Thus, ADAGA is able to produce results as close as possible to real results (obtained when no resource constraint is faced). The results obtained through experiments prove the efficiency of ADAGA.

Sharing mobile databases in dynamically configurable environments

In an environment with support for mobile computing, we may have a collection of autonomous, distributed, heterogeneous and mobile databases, denoted Mobile Database Community (MDBC), in which each database user can access databases in the community through a wireless communication infrastructure. In such an environment, new participants may join to an MDBC as they move within communication range of one or more hosts which are members of the MDBC. Furthermore, MDBC participants may transiently disconnect from the network due to communication disruptions or to save power. Therefore, an MDBC can be characterized as a dynamically configurable environment. This paper describes an agent-based architecture, denoted AMDB (Accessing Mobile Databases), which enables such communities to be formed opportunistically over mobile database hosts in ad hoc configurable environments. The AMDB architecture is fully distributed and has the capability of exploiting physical mobility of hosts and logical mobility of database queries and their results across mobile hosts.

A semantic-serializability based fully-distributed concurrency control mechanism for mobile multi-database systems

The nodes of a mobile ad hoc network (MANET) represent mobile computers in which database systems (DBSs) may reside. In such an environment, we may have a mobile multidatabase system (mobile MDBS), i.e., a collection of autonomous, distributed, heterogeneous and mobile DBSs, where each mobile computer can access multiple DBSs of that collection by means of global transactions. In this paper we propose SESAMO, which is a concurrency control mechanism that does not require that a single mobile host (a node in MANET) plays the role of the centralized coordinator for all global transactions executed in the mobile MDBS. Moreover, SESAMO allows that subtransactions commit independently of the commitment of their respective global transactions, since it is based on the semantic serializability, which relaxes global serializability.

Introducing self-adaptability into transaction processing

In a database system, the scheduler has the goal of synchronizing operations belonging to several concurrent transactions. In order to achieve its goal, the scheduler implements a concurrency control protocol, which may have either conservative or aggressive behavior. This paper presents a self-adaptable scheduler, called Intelligent Transaction Scheduler (ITS), which has the ability of dynamically changing its behavior (from conservative to aggressive and vice-versa) to adapt itself to the characteristics of the computing environment (e.g., the aborted transaction rate and conflicting operation rate). The proposed scheduler adapts its behavior without any human interference by using an expert system based on fuzzy logic. In order to evaluate ITS, it was applied in a Mobile Database Community (MDBC). An MDBC can be characterized as a dynamically configurable environment, since an MDBC is a dynamic collection of autonomous mobile databases, interconnected through a mobile ad hoc network (MANET). Therefore, self-adaptability plays a key role for schedulers running in dynamically configurable environments. The experimentation results show the efficiency of ITS for synchronizing transactions in MDBCs

On mobile transaction processing in dynamically configurable mobile database communities

The topology of a mobile ad hoc network (MANET) may change randomly and rapidly at an unpredictable time, since nodes are free to move arbitrarily. In such an environment, we may have a collection of autonomous, distributed, heterogeneous and mobile databases (denoted mobile database community (MDBC)). This paper describes an approach for controlling concurrency of mobile transactions executed in MDBCs. The proposed approach is based on the use of semantic knowledge to relax the notion of absolute transaction atomicity. Supported by this new concept of atomicity, we propose a new correctness criterion, denoted mobile semantic serializability, for the execution of concurrent transactions in MDBCs. The proposed correctness criterion provides a high degree of inter-transaction parallelism and ensures consistency of the mobile database.

Increasing mobile transaction concurrency in dynamically configurable environments

The topology of a mobile ad hoc network (MANET) may change randomly and rapidly at unpredictable time, since nodes are free to move arbitrarily. In such an environment, we may have a collection of autonomous, distributed, heterogeneous and mobile databases (denoted mobile database community or MDBC). This paper describes an approach for controlling concurrency of mobile transactions executed in MDBCs. The proposed approach is based on the use of semantic knowledge to relax the notion of absolute transaction atomicity. Supported by this new concept of atomicity, we propose a new correctness criterion, denoted mobile semantic serializability, for the execution of concurrent transactions in MDBCs. The proposed correctness criterion provides a high degree of inter-transaction parallelism and ensures consistency of the mobile database.

On the effect of human mobility to the design of metropolitan mobile opportunistic networks of sensors

Abstract We live in a world where demand for monitoring natural and artificial phenomena is growing. The practical importance of Sensor Networks is continuously increasing in our society due to their broad applicability to tasks such as traffic and air-pollution monitoring, forest-fire detection, agriculture, and battlefield communication. Furthermore, we have seen the emergence of sensor technology being integrated in everyday objects such as cars, traffic lights, bicycles, phones, and even being attached to living beings such as dolphins, trees, and humans. The consequence of this widespread use of sensors is that new sensor network infrastructures may be built out of static (e.g., traffic lights) and mobile nodes (e.g., mobile phones, cars). The use of smart devices carried by people in sensor network infrastructures creates a new paradigm we refer to as Social Networks of Sensors (SNoS). This kind of opportunistic network may be fruitful and economically advantageous where the connectivity, the performance, of the scalability provided by cellular networks fail to provide an adequate quality of service. This paper delves into the issue of understanding the impact of human mobility patterns to the performance of sensor network infrastructures with respect to four different metrics, namely: detection time, report time, data delivery rate, and network coverage area ratio. Moreover, we evaluate the impact of several other mobility patterns (in addition to human mobility) to the performance of these sensor networks on the four metrics above. Finally, we propose possible improvements to the design of sensor network infrastructures.

Evaluating the Performance of Social Networks of Sensors under Different Mobility Models

Sensor Networks are becoming ubiquitous in our society due to their broad applicability to data intensive tasks such as keeping air population to safe levels, efficient communication in military applications, to mention but a few. Furthermore, we have seen the emergence of sensor technology being integrated in everyday objects such as cars, traffic lights, phones, and even being attached to living beings such as dolphins, birds and humans. The consequence of this widespread use of sensors is that new sensor network infrastructures may be built out of static and mobile nodes. When mobility is a variable one should define which mobility model is best for the infrastructure given their differences. This paper evaluates which mobility pattern is best suited to be used in a Social Network of Sensors (SNoS). We evaluate several mobility models and measure the efficiency of information flow in a SNoS if mobile sensors follow these mobility patterns. The paper provides us with a greater understanding of the benefits of mobility in realistic scenarios.

Balancing energy consumption and memory usage in sensor data processing

Algorithms for data processing in sensor nodes of wireless sensor networks should be able to handle the resource limitations the nodes face (i.e. energy and memory). An important issue to be considered is that the materialization of large amounts of data in sensor nodes may cause memory overflows and consequently data losses. On the other hand, the excessive sending of data packets to other nodes may result in unacceptable energy consumption levels. Some alternatives to save energy and also avoid memory overflows involve in-network aggregation and reductions in sensor activities (when sensor nodes experience resource constraints). In this paper we study energy-memory tradeoffs in sensor nodes and how they affect the accuracy of query results in wireless sensor networks. We propose an adaptive data processing strategy to balance energy consumption and memory usage at sensor node level. Our goal is to maximize sensor lifetime while also maintaining the accuracy of query results. We implemented our approach by means of an algorithm called ADAGA (ADaptive AGgregation Algorithm for sensor networks), which process in-network aggregation in sensor nodes. ADAGA is able to adapt its behavior according to energy and memory availabilities by dynamically adjusting data collection and data sending intervals. The results on the efficiency of our approach are also presented at the end of the paper.

Improving Multidimensional Wireless Sensor Network Lifetime Using Pearson Correlation and Fractal Clustering

An efficient strategy for reducing message transmission in a wireless sensor network (WSN) is to group sensors by means of an abstraction denoted cluster. The key idea behind the cluster formation process is to identify a set of sensors whose sensed values present some data correlation. Nowadays, sensors are able to simultaneously sense multiple different physical phenomena, yielding in this way multidimensional data. This paper presents three methods for clustering sensors in WSNs whose sensors collect multidimensional data. The proposed approaches implement the concept of multidimensional behavioral clustering. To show the benefits introduced by the proposed methods, a prototype has been implemented and experiments have been carried out on real data. The results prove that the proposed methods decrease the amount of data flowing in the network and present low root-mean-square error (RMSE).