Earticle

To address the accuracy degradation as well as prolonged convergence time due to the inherent data heterogeneity among end-devices in federated learning (FL), we introduce the joint batch size and weighted aggregation adjustment problem, which is non-convex problem. To adjust optimal hyperparameters, we develop deep reinforcement learning (DRL) to empower a mechanism known as Batch size and Weighted aggregation Adjustment (BWA). Experimental evaluation demonstrates that BWA not only outperforms methods optimized solely from either a local training or server perspective but also achieves higher accuracy, with an increase of up to 5.53% compared to FedAvg, and additionally accelerates convergence speeds.

In this paper we propose high rate turbo code (HRTC) that exploits the potential correlation present between the data of the neighboring sensor nodes in a wireless sensor network. The problem of implosion and overlap, faced by the sensor nodes, gives rise to data redundancy in the sensor network. This paper combines source -channel coding wi th data aggregation using HRTC to reduce the redundant information produced amongst the nodes and to make the network more resourceful and reliable. This paper also includes the design of the proposed HRTC that uses deterministic interleaver to ensure good permutations.

고속도로에서 수집 저장되는 이력자료의 활용도를 효과적으로 향상시키기 위해 Archived Data User Service (ADUS)/Archived Data Management System(ADMS) 구축에 대한 연구가 활발하게 진행되고 있다. 그러나, 방대한 이력교통자료의 활용도를 높이기 위해 우선 차량검지기 자료처리과정에 대한 체계적인 조사 분석을 통한 차량검지기 자료의 신뢰성 확보가 선행되어야 한다. 본 연구에서는 고속도로 차량검지기 자료처리과정에 대한 종합적인 문제점 분석을 통해 고속도로 교통관리 시스템의 개선사항에 대하여 살펴보았으며, 차로 이용률과 관련된 문제점과 개선방향에 연구의 초점을 맞추어 수행하였다. 본 연구에서는 우선 하루 동안 전체 검지기에서 수집된 자료 중 오류데이터가 차지하는 부분을 분석하였으며, 전체 수집 자료중 15%가 신뢰성에 영향을 미치는 오류데이터로 분석되었다. 또한, 고속도로 차량검지기에서 수집되는 자료의 차로 이용률(Lane Usage)을 살펴보았다. 분석결과 2차로구간에서는 1 2차로간 평균 차로이용률이 12% 정도의 차이를 나타내는 것으로, 3차로 구간에서는 1 2차로의 37%에 비해 3차로의 차로이용률이 24%로 비교적 낮게 분포하는 것으로 분석되었다. 4차로 분석구간에서는 1 4차로가 2 3차로에 비해 낮은 차로이용률을 나타내는 것으로 분석되어 실제 차로이용률이 차로별로 동일하지 않은 분포를 가짐을 확인할 수 있었다. 고속도로 차량검지기 자료처리과정에 대한 조사 분석 결과 더 신뢰성 있는 자료를 수집 저장하기 위해서는 고속도로에서 각 차로별 차로 이용률이 시간대에 따라 다르게 분포하는 점을 고려한 검지기 지점 데이터 생성 process에 대한 추가적인 연구가 필요하다.

Traffic condition monitoring system serves as the foundation for all intelligent transportation system operation. Loop detectors and Video Image Processing are the most widely common technology approach to condition monitoring in korea Highways. Lane Usage is defined as the proportion of total link volume served by each lane. In this research, the lane Usage(LU) of two lane link for one day. Interval is 56% : 44%. The LU of three lane link is 39% : 37% : 24%. The LU of four lane link is 25% : 29% : 26% : 21%. These analysis reveal that each lane distributions of link are not same. This research investigates the general concept of lane usage by using collected loop detector data and the investigated that lane distribution is different by traffic lane and lane usage is consistent by time of day.

집계화 수준과 자료수집 주기가 상이한 시계열자료를 결합하여 짧은 주기의 일관성있는 시계열자료를 구축하는 기법을 제시하였다. 추정과 예측을 확장 회귀식의 추정으로 단일화한 Salkever(1971)의 기법을 활용하여, 자료의 일관성 확보문제를 확장 회귀식체계 하에서의 상이한 확장 회귀식에 등장하는 모수들간의 제약으로 전환할 수 있었다. 따라서, SURE형태의 확장 회귀식체계에 반복 GLS 추정기법을 적용하여 짧은 주기의 일관성있는 시계열자료를 얻을 수 있었다. 본 논문의 기법을 적용하여, 집계화 수준과 자료수집 주기가 상이한 [에너지총조사]와 [에너지통계연보]의 시계열자료를 결합하여 석유화학(대분류부문)에 속하는 3개 중분류부문에 대한 연간 시계열자료를 구축하였다.

We propose an econometric method to construct a logically- consistent time series from multiple data sources with differences in sector aggregation and survey periodicity. Logical consistency is the requirement that at each period the total of the predicted values of component subsectors should be equal to the actual value of the integrating sector. Building upon Salkever (1971), a consistent time series is constructed by applying iterative GLS to a system of augmented regression equations with cross-equation restrictions on the parameters. We successfully applied our method to the construction of annual data for the three subsectors of Petrochemicals by combining annual data (from Yearbook of Energy Statistics) on Petrochemicals and triennial data (from Energy Consumption Survey) on its three subsectors.

The issue of data aggregation in wireless sensor networks was studied. Data aggregation was an efficient in-network data processing method. It reduced data redundancy and improved information quality, which may save communication energy and increase collection efficiency. In this dissertation, data aggregation solution (DAS) was proposed and realized. DAS used a cluster structure based network and it had three parts. Firstly, the cluster head would not submit any packets derived from the cluster members. Instead, it fused them into an outcome packet and then sent it to the sink node. Secondly, the cluster head scheduled the nodes with low energy level into sleep. Thirdly, the cluster head would use a combined forecasting algorithm to estimate the data of sleeping nodes. Simulation tests were carried out, and simulation results showed that DAS had good performance. It not only extended the network lifetime greatly, but also provided better assurance of data quality.

Energy consumption is one of the most challenging constraintsof the design and implementation of the sensor network. Underwater sensor networking is the technology that enables the applications like environment monitoring, underwater exploration, seismic monitoring and other surveillance applications. In underwater sensor network,a sensor node senses the data and transmits it to the sink. Many routing algorithms have been proposed in order to make the network phase of UWSNs more efficient.In this report, we present a review and comparison of various data collection algorithms and clustering schemes, proposed recently in order to execute the demands of the ongoing researches. The main goal of data aggregation technique is to accumulate data in an energy efficient manner for a long-term network monitoring. The main purpose of this study is to present algorithms addressing issues like deployment and localization in UWSNs under different conditions.

We aim to use public data different from the remote BEMS energy diagnostics technology and already established and then switch the conventional operation environment to a big-data-based integrated management environment to operate and build a building energy management environment of maximized efficiency. In Step 1, various network management environments of the system integrated with a big data platform and the BEMS management system are used to collect logs created in various types of data by means of the big data platform. In Step 2, the collected data are stored in the HDFS (Hadoop Distributed File System) to manage the data in real time about internal and external changes on the basis of integration analysis, for example, relations and interrelation for automatic efficient management.

There exist two key problems about data aggregation that should be thoroughly explored - algorithm design in networking layer, and algorithm design in application layer. Those two problems should be subtlety tackled in termers of high efficiency and robustness. Therefore, the former one requires the survivability and highly reliable design at networking layer, the latter one usually asks for high efficiency and robustness at application layer. Moreover, the optimization of algorithms is also considered for further enhancement. The integrity check is a key requirement for optimization. The context-aware and cross-layer design is applied in the optimization. A dynamic fragment odd-even parity checking code is proposed, and a context-aware aggregative integrity check code is proposed.

A Vehicular Ad-Hoc Network (VANET) is categorized as a Mobile Ad-Hoc Network (MANET) which delivers wireless network servies with an aim to improve road safety and enhance driving comfort. Diverse applications of Vehicular Ad-Hoc Networks such as infotainment, road safety and public safety have made VANETs as a notable and emerging area of research and development. As of now, numerous vehicular ad-hoc network research projects have been mainly aimed at data security and routing. This has raised a critical problem of data congestion and loss of data accuracy in VANETs. A major challenge in VANETs is to provide efficient data communication and propogation for precise and valuable information. This paper presents a generalized framework for message aggregation. Message Aggregation can be used to transmit minimal data and to enhance the communication efficiency thus reducing the communication overhead in VANETs. This will help in reducing the redundancy in VANETs resulting in dissemination of precise information

Sensor networks have a significant potential in diverse applications, and some have already been deployed in monitoring system of living environment. With the increasing complexity of application logic, difficulties in monitoring sensor networks have become a barrier to the adoption of these networks. The difficulties are due not only to their inherently distributed nature but also to the need for mechanisms to address their harsh operating conditions such as unreliable communications, faulty nodes, and extremely constrained resources. Living Environment monitoring is composed mainly of sensor data on air, water, and ecotourism quality. Wireless sensor networks(WSNs) entail a substantial loss of energy because there is a need for some mechanisms that can select multiple communications in single communications. This kind of merging is called data aggregation. This paper presents a secure and authentication-based approach to the data aggregation of living environment. User authentication is performed using a secure hash algorithm. This paper also introduces a neural network to check bad packet communications over the network. The results indicate that the proposed approach is more reliable and efficient than existing ones.

This paper present a review on the data aggregation techniques in WSNs. Data aggregation techniques have become an major part of WSNs to overcome the problem of flooding at the base station. Also data aggregation techniques reduce the communication cost of WSNs; because it reduces the redundant data by aggregation. Various data aggregation techniques have been considered in this paper. The data aggregation results in reduction of redundant data therefore reduce the amount of data transmission packets size which will reduce the energy consumption in each round. This paper ends with the shortcoming of data aggregation techniques.

Wireless multimedia sensor network (WMSN) is an emerging type of sensor network that contains sensor nodes equipped with cameras, microphones, and other sensors producing multimedia content. These sensors require sufficient energy to acquire, transmit, and process heterogeneous and relatively large volume of raw data. Due to resource restrictions of wireless sensor nodes, it is important to minimize the amount of redundant data transmission so that the average sensor lifetime and the overall bandwidth utilization are improved. Data aggregation is vital in conserving energy by eliminating the inherent redundancy of raw data in WMSNs. While WMSNs incur high maintenance overhead in dynamic scenarios, WMSNs should be capable of timely fulfilling its mission without losing important information in event-based applications. This paper focuses on designing a tree-based data aggregation scheme. Using simulations, we investigate the performance of the data aggregation in terms of aggregation gain and energy consumption for WMSNs.

Data Aggregation Scheduling in Ubiquitous sensor networks is a major research interest for many researchers with the objective of reducing energy consumption and maximizing the network lifetime. Also the capacity of a channel for transmitting quantum states called the quantum capacity should be maximized,The sensor nodes should periodically forward data to the base station. Since battery power and bandwidth are the two resources to be effectively utilized the sensor nodes forwards data to the nearest neighbor node which inturn forwards the data to its neighbors and this repeats until all the sensors data aggregates at the base station. In a large network scheduling the sensor nodes for forwarding the data is very much essential and since the topology of the USN changes dynamically , finding such a schedule is also very difficult. In [1] the authors proposed a bosonic network model for communication. In this paper we present a quantum bosonic data aggregation method for constructing the optimal schedule of nodes for data aggregation called QBDAS (Quantum Boson Data aggregation scheduling) protocol. The main idea is to search a optimal schedule through utilizing the observable or measurable physical properties the network topology.

Because a sensor node has limited resources, such as battery capacity, data aggregation techniques have been proposed for wireless sensor networks (WSNs). On the other hand, the provision of efficient data aggregation for preserving data privacy is challenging issue in WSNs. Existing data aggregation methods for preserving data privacy are CPDA, SMART, Twin-Key based method, and GP2S. However, they have a main limitation that communication cost for network construction is considerably high. To resolve the problem, we propose a privacy preserving data aggregation scheme based on Hilbert curve for WSNs. For data aggregation, we utilize a tree-based network structure which minimizes communication among sibling sensor nodes for network construction. Moreover, we adapt a Hilbert curve technique to preserve data privacy. Because the sending data is encrypted by using a unique Hilbert value, it is very difficult to trace a real value even though attackers overhear the sending data. Through our performance analysis, we show that our data aggregation scheme outperforms the existing methods in terms of energy efficiency and privacy preservation

To meet the demands of wireless sensor networks (WSNs) where data are usually aggregated at a single source prior to transmitting to any distant user, there is a need to establish a tree structure inside to aggregate data. In this paper, an adaptive energy aware data aggregation tree (AEDT) is proposed. The proposed tree uses the maximum energy available node as the data aggregator node. The tree incorporates sleep and awake technology where the communicating node and the parent node are only in awake state rest all the nodes go to sleep state saving the network energy and enhancing the network lifetime. When the traffic load crosses the threshold value, then the packets are accepted adaptively according to the communication capacity of the parent node. The proposed tree maintains a memory table which stores the value of each selected path. Path selection is based on shortest path algorithm where the node with highest available energy is always selected as forwarding node. By simulation results, we show that our proposed tree enhances network lifetime minimizes energy consumption and achieves good delivery ratio with reduced delay.

In scenarios of real-time data collection in long-term deployed Wireless Sensor Networks (WSNs), low-latency data collection and long network lifetime become key issue. We propose a Long-Lifetime and Low-Latency Data Aggregation Scheduling algorithm (L4DAS) in wireless sensor networks. Firstly, we formally formulate the problem of long-lifetime and minimum-latency aggregation scheduling as a constrained optimization problem, and then propose an approximation algorithm for this problem by constructing a degree-bounded minimum height spanning tree as aggregation tree and designing a maximum interference priority scheduling scheme to schedule the transmission of nodes in aggregation tree. Finally, through the simulation and comparisons, we prove the effectiveness of the algorithm.

Wireless sensor network consists of a huge quantity of less-price sensor nodes. These nodes has restricted power of battery, and the replacement of battery is not a simple task in wireless sensor networks because there are a huge quantity of nodes. Data Aggregation is a significant method to attain power efficiency in wireless sensor network. Data aggregation at the sink by all the nodes results in flooding of the data which causes greatest energy utilization. Though a lot of protocols are planned so far to get better the energy efficiency further but still a lot improvement can be made. In this paper, various data aggregation techniques have been discussed. The overall purpose of this survey is to explain data aggregation techniques and to find limitations of General Self-Organized Tree-Based Energy-Balance Routing Protocol (GSTEB).

As modern agriculture develops, the demand to the agricultural information acquirement technology grows. This paper applies the wireless sensing data acquisition technology to the agricultural production. A data aggregation method based on matching pursuit algorithm is proposed according to the features and function of WSN data aggregation. By the theoretical analysis and experimental simulation, it’s proved that the data aggregation of wireless sensor network can be achieved with the application of matching pursuit algorithm. Thus both the amount of data transmission and energy consumption of the network are reduced effectively, the life of the network extends. And the high data reconstruction accuracy helps the large-scale application.

This paper presents an efficient clustering scheme for data aggregation considering mobility to reduce the network lifetime in Mobile Wireless Sensor Networks (MWSNs). In MWSNs, an accurate data aggregation is affected by failure in time-critical data transmission due to the differences of mobility between cluster head and its members. Consequently, selecting a node with dissimilar mobility from members as a cluster head can lead to unstable clustering. Hence, we propose a clustering scheme considering the mobility and energy to minimize the number of nodes that moves away from the current cluster head before the next cluster formation. The proposed scheme is processed in two-stages. During the first phase of clustering, all nodes calculate their potential score based on the similarity of movement, residual energy and density in distributed manner. Each node decides whether the node itself should be a cluster head or not using a potential score. The second stages, each node select its cluster head among those cluster head candidates considering the link connection time and an amount of energy to transmit the collected data. A higher residual energy node with the smallest movement deviation is selected as a cluster head. In addition to, we propose a mobility adaptive slot utilization to reduce the idle slot. The experimental result shows that the proposed scheme can provide an efficient data aggregation in terms of network lifetime and the number of nodes leaving its cluster head.