As we know, the optimization of large-scale and complex wireless networks is a fundamental research problem. However, deriving an optimal solution is challenging due to the following two facts; (i) the solution space is dependent of the scale of the wireless environment, namely the number of terminals and that of base station (BS) candidates, and (ii) when considering a dynamic environment that changes over time rather than a static one, the solution space grows proportionally to the time scale. One promising approach to the acceleration in searching an optimal solution is to reduce the solution space. To this end, we propose a graph-based reduction approach for dynamic wireless environments. Namely, we represent the dynamic wireless environment as a time-series graph and compress it. The key idea of our proposal is twofold. Firstly, we describe the relationship between terminals and BS candidates as a bipartite graph. Secondly, we reduce the time-series graph in temporal and spatial dimensions by iteratively merging graph pairs and node pairs. As an application of our graph compression, we solve a BS placement problem, which is a well-known optimization problem, using an extended greedy algorithm. Moreover, we discuss the effect of the graph compression on the solution quality through experiments.

Generally, performing network simulations and deep learning using graph structure such as GNN (Graph Neural Network) requires network topologies. However, it is undesirable to directly give real large-scale networks to input of such computation tasks since the computational complexity is mainly dominated by the network size. One of promising solutions to alleviate this problem is graph reduction which reduces the size of the network, i.e., the number of nodes, while preserving its structural properties. Previous researches on the graph reduction have mainly focused on the extent to which structural property of the network is preserved with respect to the reduction in network size; this means that it is questionable that the graph reduction is actually useful at the application-level. Therefore, in this paper, we focus on evaluating the robustness of networks, which is one of typical topics in the field of network science, and evaluate whether graphs reduced with various sampling strategies and coarsening algorithms maintain the robustness of the original graph in terms the size of the largest connected component. As a consequence, we reveal, for instance, that even though the graph size is halved, reduced graphs can maintain the robustness of an original graph, in particular, random graph.

In recent years, it is expected that ICDTN (Information-Centric Delay/Disruption- Tolerant Net-working) incorporating the communication paradigm of Information-Centric Networks will be realized in an environment where communication links between nodes are intermittent, and its effectiveness has been actively investigated. To realize efficient content delivery in ICDTN, it is necessary to appropriately select content request message routing and content response message routing in a network environment where heterogeneous communication links with different characteristics are intermittent. In this paper, we first clarify how communication link heterogeneity affects the communication characteristics of content routing in ICDTN. Specifically, the heterogeneity of communication links is modeled as two types of ON/OFF models with different link avail-ability. In addition, we analytically derive the average content delivery delay when the end-to-end routing is used as the routing method for request messages and the traceback routing is used as the routing method for response messages. Furthermore, through several numerical examples, we investigate the effect of the heterogeneity of communication links on the average content delivery delay.

Acquisition and estimation of the topology of evolving and large-scale networks such as communication networks and social networks are not trivial because of their scale, complexity, and dynamics. In general, the topology of a communication network can be represented as a graph composed of many vertices and edges, and the estimation problem of the network topology can be handled as a topology estimation problem of the topology from limited knowledge on the graph. The network topology estimation problem covers a wide range of variations depending on the available data, constraints, and the objective function. Variants of the network topology estimation problem can be classified into two categories: direct and indirect. In the indirect network topology estimation problem, only information regarding the network topology to be estimated is known. In this paper, we propose an indirect network topology estimation method called TOPFLOW (network TOPology inference from FLOW sets), which estimates the topology of the entire network from the limited number of flow sets observed at measurement nodes in the network. Furthermore, we extensively investigate the effectiveness of TOPFLOW through a number of experiments with diverse networks with different structures and scales. Our findings include that the estimation accuracy grows almost linearly as the ratio of measurement nodes increases in some network topologies.

In recent years, a statistical approach called sparse modeling has been studied extensively for estimating unobserved model parameters from a small number of observations using the sparsity of model parameters. Although sparse modeling has been applied to many practical problems in the fields of signal processing and image processing, to the best of our knowledge, few studies have applied it to the field of information networking. In this paper, we investigate whether a sparse representation of network topology can be obtained from a dictionary trained with a dictionary learning algorithm in sparse modeling. Specifically, we train a dictionary from a number of learning network topologies using the K-SVD algorithm, which is one of conventional dictionary learning algorithms, and obtain a sparse representation of the network topology by solving an l0-norm minimization problem for given network topology and the trained dictionary. Furthermore, through experiments, the effects of several factors - the network (i.e., topology and network size) and the dictionary (i.e., dictionary size) - on sparse representation of network topologies are investigated. Our finding includes that graphs whose structure is uniform (e.g., tree) and networks with cluster structure are suitable for sparse representation of network topologies.

In recent years, a statistical approach called sparse modeling has been studied extensively for estimating unobserved model parameters from a small number of observations by using the sparsity of model parameters. Although sparse modeling has been applied to many practical problems in the fields of signal processing and image processing, to the best of our knowledge, few studies have applied it to the field of information networking. In this paper, we investigate how sparse modeling can be applied to a network flow problem. Specifically, we focus on the minimum link flow problem that is similar to the classical minimum cost flow problem except that its objective is to minimize the number of links consisting a flow rather than the total link cost. We present a sparsemodeling- based formulation of the minimum link flow problem and investigate how effectively our formulation of the minimum link flow problem can be solved using a conventional greedy algorithm called Orthogonal Matching Pursuit (OMP). We also extend our sparse-modeling-based approach to a constrained minimum link flow problem with finite link capacities. For solving the constrained minimum link flow problem, we propose a greedy algorithm called Constrained Orthogonal Matching Pursuit (COMP).

In recent years, a statistical approach for estimating unobserved model parameters from a small number of observations utilizing the sparsity of model parameters called sparse modeling have been extensively studied. In our previous work, we have shown the effectiveness of sparse modeling for a network flow problem called minimum link flow problem, which finds, for given incoming/outgoing rate requirements at nodes, a set of flows satisfying requirements with the least number of links. This paper extends our sparse-modeling based approach to a more complex problem - estimation of betweenness centrality, which is one of the major graph indices. In this paper, we present a sparse-modeling based solution for betweenness centrality estimation. Betweenness centralities of all nodes in an undirected graph are estimated from shortest-path trees, each of which is obtained as the solution for the l1-norm minimization problem.

In this video, we propose a new interaction method using mobile devices in Intelligent Space (iSpace). This interaction, called R-Fii (Real-world Flexible Interaction Interface), uses mobile devices, mainly cellular phones, as stable interface channels to iSpace. The users are able to control any objects and get information of the objects in iSpace with R-Fii.