China
Africa
Explore chapters and articles related to this topic
Dynamic Risk Management
Published in Nicholas Kolokotronis, Stavros Shiaeles, Cyber-Security Threats, Actors, and Dynamic Mitigation, 2021
Ioannis Koufos, Nicholas Kolokotronis, Konstantinos Limniotis
Scalable Bayesian interference in BAGs could be achieved with the help of the sum-product algorithm also known as Belief Propagation (BP) algorithm. BP reduces the computation of unconditional probabilities in a Bayesian network but requires the graph to be a tree/poly-tree [25]. However, various extensions of the sum-product algorithm could be incorporated in graphs. Aforementioned algorithms require the conversion of the Bayesian attack graph to a factor graph. While Bayesian attack graphs allow calculating the joint probability distribution as a product of factors, factor graphs on the other hand allow the factor decomposition into subsets. The joint probability distribution is then computed as the product of all subsets PrX1,…,Xn=∏i=1mfiXi
Signature Generation Algorithms for Polymorphic Worms
Published in Mohssen Mohammed, Al-Sakib Khan Pathan, Automatic Defense Against Zero-day Polymorphic Worms in Communication Networks, 2016
Mohssen Mohammed, Al-Sakib Khan Pathan
These are usually denoted as open circles and filled dots (Figure 7.14, center). A factor graph is like an undirected model, which represents a factorization of the joint probability distribution: Each factor is a nonnegative function of the variables connected to the corresponding factor node. Thus, for the factor graph in Figure 7.14, center, we have P(A,B,C,D,E)=cg1(A,C)g2(B,C)g3(B,D),g4(C,D)g5(C,E)g6(D,E)
Factor Graphs and Message Passing Algorithms
Published in Erchin Serpedin, Thomas Chen, Dinesh Rajan, Mathematical Foundations for SIGNAL PROCESSING, COMMUNICATIONS, AND NETWORKING, 2012
Aitzaz Ahmad, Erchin Serpedin, Khalid A. Qaraqe
Factor graphs find numerous applications in modeling and inference in systems. Some of these applications include iterative decoding of turbo codes [17], autoregressive model parameter estimation [18], electromyographic (EMG) signal separation [19], receiver design [20] and joint iterative detection and decoding [21]. Factor graphs have also been employed in linear equalization [22], LMMSE turbo equalization [23] and adaptive ualization [24]. Iterative detection schemes for ISI channels based on the sum-product inference algorithm in factor graphs representing the joint a posteriori probability of transmitted symbols are proposed in [25]. Factor graphs have also found applications in joint decoding and phase estimation [26], [27]. Since factor graphs offer an opportunity for modeling and inference simultaneously, they are also finding increasing applications in channel modeling and statistical inference.
Performance analysis of alternating minimization based low complexity detection for MIMO communication system
Published in Automatika, 2023
Kasiselvanathan M., Manikandan Rajagopal, Teresa V. V., Prabhakar Krishnan
Hybrid BP (Belief Propagation) and EP (Expectation Propagation) receivers, initially addressed BP algorithm’s convergence issue using auxiliary variables in factor graph-based near-optimal iterative receivers [28]. Jiang et al. [29] suggested quick processing approaches with low complexity and provided iterative receivers to comprehend linear inverse matrices issues. The iterative technique updated the process separately on a small-size block by using the block matrix attributes. MIMO detection with minimal complexity utilising adaptive mitigation is introduced by Park et al. [30]. Imperfect precoded matrices were used by quantization error-based downlinked multiuser MIMO to lessen interferences and receive required signals at receivers. To achieve reduced complexities, Zhao and Du [31] suggested LRA (lattice reduction assisted) based MIMO detections.
Key web search algorithm based on service ontology
Published in International Journal of Computers and Applications, 2021
In this task, we use the graph of one-factor graph for representation, which is the undirected bigraph suitable to process the optimization task. The factor graph is bigraph, composed of two different types of nodes, variable node and function node respectively, as is shown in Figure 2.
Real-time collision handling in railway transport network: an agent-based modeling and simulation approach
Published in Transportation Letters, 2019
Poulami Dalapati, Abhijeet Padhy, Bhawana Mishra, Animesh Dutta, Swapan Bhattacharya
As described in Subsection ‘Collision detection in a real-time scenario’, the problem is very challenging in real-time scenario. The conflict situation can be prevented in a distributed manner through message-passing among neighboring trains, stations, and junctions within the communication range. In order to solve such problem, we represent each train, station, and junction as an autonomous agent. These agents are capable of communicating and coordinating with their neighbors through message passing. We use the notion of max-sum algorithm for decentralized coordination Farinelli et al. (2008), Stranders et al. (2009) to solve such problem. Here, agents negotiate by exchanging messages locally to achieve a desired solution globally. Within the communication range, all trains can communicate with each other and always take part in the collision detection–resolution task. First the agents perform collision detection. In this phase, all the agents check for a situation when the distance between two trains is less than their actual headway distance, which may lead to a fatal collision. If such scenario arises then collision resolution technique is needed to prevent the mishap. Collision resolution is based on agent cooperation and negotiation. During collision detection phase, all the agents involved in collision scenario check for all the metric parameters: headway distance, braking distance, critical distance and also their possible decision states. If more than one potential threats are detected, then the most fatal collision is handled first and the lower one is handled later. The participating agents search for the safe state from all possible set of states using the idea of max-sum algorithm as discussed (Farinelli et al. 2008; Stranders et al. 2009) in this section to detect collision. The proposed algorithm works iteratively until a feasible solution is found. In each iteration, all the agents exchange their new modified state, generated by max-sum approach, with their neighboring agents through message passing. In this paper, two possible states have been taken: move further and stop applying full brake. If there are several alternatives for the agents, then the best possible solution is chosen depending on trains priority, minimum braking distance, and critical distance. Finally all the agents acquire decided action of state and send the messages to the nearby train agents, station agents, and junction agents. We first represent each agent as a function U (utility) and a variable (state), which are the vertices of the factor graph (Kschischang et al. 2001). The utility of any agent depends on its own state and the state of its neighbors, where, and z is the total number of factors. So, the function node is connected with its own variable node and the variable nodes of its neighbors.