Nowadays, wireless sensor networks are widely used in many environmental monitoring applications. However, due to the limitation of the current hardware technology, sensors are often battery powered and it is very difficult to change batteries. Therefore, for applications over wireless sensor networks, it is a critical issue to save the energy of sensors. Many attempts have been made to answer various types of queries energy-efficiently, such as max, top-k, and skylines. However, all of them return the readings of individual sensors that satisfy the query constraint. In practice, query results based on individual sensor readings are unreliable because sensor readings are often noisy. Thus, in this paper, we present a new type of query, max aggregate query over a region (i.e. max regional aggregate), which aims to find a fixed-size region whose regional aggregate is the maximum among all the possible regions with the same size. Compared to traditional max queries, max regional aggregate is more reliable in detecting events. Designing an energy-efficient approach to answer max regional aggregate is non-trivial because a huge number of regions need to be investigated. Thus, in this paper, we propose a novel two-level sampling approach, with region and sensor sampling, to collect the sensor readings intelligently and compute the approximate max regional aggregate based on the collected results. Specifically, region sampling is used to select regions and sensor sampling isused to choose the sensors within a selected region. Our extensive simulation results demonstrate that the proposed two-level sampling approach can answer the max regional aggregate energy-efficiently with a desirable accuracy.

Nowadays, wireless sensor networks have been widely used in many monitoring applications. Due to the low quality of sensors and random effects of the environments, however, it is well known that the collected sensor data are noisy. Therefore, it is very critical to clean the sensor data before using them to answer queries or conduct data analysis. Popular data cleaning approaches, such as the moving average, cannot meet the requirements of both energy efficiency and quick response time in many sensor related applications. In this paper, we propose a hybrid sensor data cleaning approach with confidence. Specifically, we propose a smart weighted moving average (WMA) algorithm that collects confidence data from sensors and computes the weighted moving average. The rationale behind the WMA algorithm is to draw more samples for a particular value that is of great importance to the moving average, and provide higher confidence weight for this value, such that this important value can be quickly reflected in the moving average. Based on our extensive simulation results, we demonstrate that, compared to the simple moving average (SMA), our WMA approach can effectively clean data and offer quick response time.

The newly emerging wireless backhaul network has fundamental difficulties in supporting Voice over IP (VoIP) applications due to the MAC overheads introduced by huge amounts of small packets. Packet aggregation is a promising approach to mitigate these overheads. However, previous approaches to such problems are often stringent, not adaptive to the change of channel conditions. They are operated by each TAP (Transit Access Point) separately without any coordination in the use of shared channels. As a result, they fail to ensure the VoIP quality in terms of delay and loss. The major contribution of this paper is the proposal of a coordinated aggregation algorithm, which is adaptive and distributed. By coordinating with neighboring TAPs, the proposed algorithm is able to assign an appropriate aggregation rate to each TAP, aiming at better channel utilization and lower packet loss and delay. We evaluate this design by comprehensive analysis and simulations. The simulation results show that our algorithm significantly improves the VoIP capacity in wireless backhaul networks and outperforms existing aggregation algorithms.

Memory-intensive applications often suffer from the poor performance of disk swapping when memory is inadequate. Remote memory sharing schemes, which provide a remote memory that is faster than the local hard disk, are able to improve the performance of such applications. Due to the limitation of being applicable within single clusters only, however, most of the previous remote memory mechanisms, such as the network memory scheme, fail to be extendable into a large scale, distributed, heterogeneous, and dynamic environment. In this work, we propose a service-oriented grid memory sharing scheme, distributed paging RAM grid (DPRG). We study the properties and criteria of large scale memory sharing, and then design major operations and optimizations to fit the usage of grid systems. We collect trace from our grid environment, and evaluate DPRG through comprehensive trace-driven simulations. Results show that DPRG significantly outperforms existing remote memory sharing schemes and supports grid computing applications effectively