Earticle

Sensor network has lot applications in the early warning and assistant of disaster environment such as debris flows, floods and forest fires. However, such disaster environment pose an interesting challenge for data collection since sensor nodes may be destroyed unpredictably and centrally, resulting in the decrease of data persistence in the network. Growth Codes Protocol (GCP) first focuses on increase sensor network data persistent in the disaster. However, the completely random data transmission way in GCP may cause a large number of invalid data transmissions and therefore, the efficiency of data collection of the protocol is not ideal in the late stage of data collection. In this paper, we propose an efficient data collection protocol (DGCP) to maximize sensor network data persistence by changing the completely random data transmission way. Packet classification mechanism and a novel dynamic probability model of data transmission in DGCP are proposed to control the effective direction of data flow. Furthermore, we found that the parameter optimization problem of the probabilistic model is a problem of searching the optimal solution in a mathematical view. Based on this property, we propose a genetic algorithm to optimize the dynamic probability model. The performance of the proposed DGCP is shown by a comparative experimental study. When compared with GCP, our DGCP has better performance in a variety of environments

컴퓨팅 패러다임이 클라우드 컴퓨팅으로 변화됨에 따라 보안 이슈가 더욱 더 중요하게 되었다. 컴퓨팅 플랫폼 서비스 제공자들은 Firewall, Identity Management 등을 제공하지만 클라우드 컴퓨팅 인프라는 사용자가 맘대로 제어하거나 새로운 장치들을 부착하여 사용할 수 없기 때문에 비교적 보안에 취약한 것이 사실이다. 이런 환경에서는 사용자 스스로 보안을 대비해야 하기 때문에 직관적인 방법으로 전체 네트워크 트래픽 상황을 가시적으로 조망할 수 있는 기법이 필요하다. 이를 위해서는 네트워크 패킷을 실시간으로 저장하고, 저장된 데이터를 준 실시간으로 분류할 수 있는 기술이 요구된다. 네트워크 패킷 분류에서 중요한 사항은 패킷 분류를 비지도 방식으로 사람의 개입 없이도 판단 기준을 지능적으로 생성하고 이를 통해 패킷을 스스로 판별할 수 있는 기술개발이 필요하다. 또한, 이를 위해서 Naive-Bayesian Classifier, Packet Chunking 등의 방법들을 활용해 사용자 개입없이 분류에 필요한 시그니쳐(Signature)를 탐색하고 이를 학습해 스스로 자동화된 패킷 분류를 실현할 수 있는 방안을 제시한다.

Paradigm shift to cloud computing has increased the importance of security. Even though public cloud computing providers such as Amazon, already provides security related service like firewall and identity management services, it is not suitable to protect data in cloud environments. Because in public cloud computing environments do not allow to use client's own security solution nor equipments. In this environments, user are supposed to do something to enhance security by their hands, so the needs of visualized security management arises. To implement visualized security management, developing near realtime data handling & packet classification mechanisms are crucial. The key technical challenges in packet classification is how to classify packet in the manner of unsupervised way without human interactions. To achieve the goal, this paper presents automated packet classification mechanism based on naive-bayesian and packet Chunking techniques, which can identify signature and does machine learning by itself without human intervention.