Lossy Aggregation Research Articles

On effective data aggregation techniques in host-based intrusion detection in MANET

A Mobile Ad Hoc Network (MANET) has been used in both commercial and tactical applications and faces security risks. Conducting cyber–attack monitoring and detection in MANET is challenging owning to limited resources. To deal with this issue, in this paper we develop lossless and lossy aggregation techniques to reduce the resources (e.g., energy consumption and bandwidth use) for transmitting security information through MANET while preserving the desired detection accuracy for cyber security operation. Particularly, our novel compression–based data aggregation technique effectively removes the duplicated information and compresses the pivotal data. The event–based data aggregation consolidates the data with the same type of predefined events. We also develop lossy data aggregation technique, namely feature–based data aggregation, which defines a series of basic features. We conduct real–world experiments and simulations to evaluate the effectiveness of our proposed data aggregation techniques in terms of energy consumption and detection accuracy.

An Adaptive Data Aggregation Algorithm in Wireless Sensor Network with Bursty Source

The Wireless Sensor network is distributed event based systems that differ from conventional communica-tion network. Sensor network has severe energy constraints, redundant low data rate, and many-to-one flows. Aggregation is a technique to avoid redundant information to save energy and other resources. There are two types of aggregations. In one of the aggregation many sensor data are embedded into single packet, thus avoiding the unnecessary packet headers, this is called lossless aggregation. In the second case the sensor data goes under statistical process (average, maximum, minimum) and results are communicated to the base station, this is called lossy aggregation, because we cannot recover the original sensor data from the received aggregated packet. The number of sensor data to be aggregated in a single packet is known as degree of ag-gregation. The main contribution of this paper is to propose an algorithm which is adaptive to choose one of the aggregations based on scenarios and degree of aggregation based on traffic. We are also suggesting a suitable buffer management to offer best Quality of Service. Our initial experiment with NS-2 implementa-tion shows significant energy savings by reducing the number of packets optimally at any given moment of time.