Frank Yeong-Sung Lin

A near-optimal sensor placement algorithm to achieve complete coverage-discrimination in sensor networks

In this letter, we develop a robust and scalable algorithm to cope with the sensor placement problem for target location under constraints of the cost limitation and the complete coverage. The problem is NP-complete for arbitrary sensor fields. The grid-based placement scenario is adopted and the sensor placement problem formulated as a combinatorial optimization problem for minimizing the maximum distance error in a sensor field under the constraints. The proposed algorithm is based on the simulated annealing approach. The experimental results reveal that, for small sensor fields, the algorithm can find the optimal sensor placement under the minimum cost limitation. Moreover, it can also find a placement with minimum distance error for large sensor fields under the cost limitation.

Relay Station Placement Strategy in IEEE 802.16j WiMAX Networks

In this paper, we study the relay station (RS) placement strategy in IEEE 802.16j WiMAX networks. Specifically, the impact of RS placement on IEEE 802.16j network performance is analyzed. A throughput maximization RS placement problem is mathematically formulated as a binary integer programming problem. We prove the NP-hardness of the formulated problem. To find the sub-optimal solution to the problem with huge input size, we propose an efficient near-optimal placement solution for IEEE 802.16j WiMAX networks. Simulations on the IEEE 802.16j network performance with our RS placement strategy are conducted. The throughput performance shows that with the deployment strategy we proposed, the IEEE 802.16j network capacity can be tremendously enhanced, especially when hotspots are present in the network.

Virtual path assignment and virtual circuit routing in ATM networks

The use of virtual paths in ATM networks reduces the call set-up delays, simplifies the hardware in the transit nodes and provides simple virtual circuit admission control. However, it also reduces the degree of capacity sharing and thus, increases the call blocking rate. We consider the following problem: given a network topology, link capacity of each physical link and traffic requirement of each origin-destination pair, we want to jointly determine the following four design variables: (1) the pairs of nodes that should have virtual paths, (2) the route of each virtual path, (3) the bandwidth assigned to each virtual path, and (4) the routing assignment for each virtual circuit (call), to minimize the expected call blocking rate subject to call set-up time constraints, or, alternately, to minimize the call set-up delay subject to the expected call blocking rate constraints. The problem is formulated as a nonlinear nondifferentiable combinatorial optimization problem. We also present simplified formulations for networks with abundant capacity or with limited capacity (two special cases of the general problem). Algorithms for solving the two special cases are proposed and implemented. We present computational results and make comparisons of different schemes.<<ETX>>

Optimal real-time admission control algorithms for the video-on-demand (VOD) service

In order to meet the quality-of-service (QOS) requirements of the VOD (video-on-demand) service, and, on the other hand, to maximize the system throughput (revenue), it is essential that the admission control algorithm be carefully designed. Two new types of admission control schemes for the VOD service are proposed. They are the enhanced strict admission control (ESAC) and the probabilistic admission control (PAC). In the ESAC schemes, we propose to use more statistics (of small amount and easily pre-calculated) than the peak frame size of the stored video information to strictly guarantee the QOS requirement and to achieve potentially much higher throughput. In the PAC schemes, we propose to use similar statistics as used in the ESAC schemes to achieve even higher throughput at the cost of some small and controllable likelihood of lost/overdue data. The admission control problems are formulated as feasibility problems where different systems of simultaneous equations are considered. For each admission control scheme, if the corresponding system of simultaneous equations has a feasible solution, then admit the call request; otherwise, reject the call. Special structures of the systems are identified so as to facilitate the development of optimal real-time admission control algorithms. Efficient optimal algorithms are also proposed to calculate the minimal buffer requirement for a given performance objective.

A simulated annealing algorithm to support the sensor placement for target location

In this paper, we develop an algorithm to cope with the sensor placement problem for target location under constraints of cost limitation and complete coverage. We adopt the grid-based placement scenario that deploys exact one sensor in one grid point at most. A target in a grid point can be positioned by a set of sensors whose transmission radius covers the grid point. The optimal sensor placement for target location is to find a sensor deployment such that targets can be positioned in any grid point of the sensor fields. However, due to the cost limitation, the optimal sensor deployment cannot be achieved frequently. Consequently, the positioning accuracy is the major issue of the problem. The distance error is one of the most natural criteria to measure the positioning accuracy. In this paper, the distance error of two indistinguishable grid points is defined as the Euclidean distance between them. We formulate the sensor placement problem as a combinatorial optimization problem for minimizing the maximum distance error in the sensor field under constraints. The sensor placement problem is NP-complete for arbitrary sensor fields. We present an efficient algorithm that is based on the simulated annealing approach to address the problem. We first compare our algorithm with the brute force approach in the case of smaller sensor fields. The evidence indicates that our algorithm can find the optimal sensor placement under the minimum cost limitation. Moreover, the simulation results also show that the proposed algorithm is very superior in terms of the positioning accuracy even in the case of larger sensor fields.

A simulated annealing algorithm for energy-efficient sensor network design

In this paper, we develop an algorithm to deploy an energy efficient sensor network such that it provides surveillance and target-positioning services. We consider to place K independent sets of sensors on a sensor field. These sets monitor the field in turn and work together when intrusion events occur. The lifetime of the sensor network is therefore prolonged up to K times. The problem is therefore a variant of the set K-cover problem, which is NP-complete. We formulate such sensor deployment problem, as a 0/1 integer programming problem. A simulated annealing based heuristic then is proposed for solving the optimization problem. The experimental results show that the proposed algorithm indicates a significant improvement in the sensor lifetime compared to the intuitive approach. Furthermore, the proposed algorithm is highly effective and efficient in terms of the overall deployment cost.

A real-time distributed routing and admission control algorithm for ATM networks

The problem of determining admission controls and a path for each admitted user pair (session) to satisfy user quality of service (QOS) requirements (required throughput, tolerable average cell delay, and tolerable cell loss probability) is considered. The problem is formulated as a nonlinear combinatorial optimization problem. The objective is the maximization of the total reward for the admitted sessions where the reward for each session reflects its priority. The emphasis is on developing a real-time distributed algorithm to determine path assignments and admission controls. In computational experiments, the proposed distributed algorithm is compared with the minimum hop algorithm on test networks with up to 61 nodes and 10000 macro sessions. The proposed algorithm achieved a 36% (on the average) improvement in the total reward over a minimum hop routing algorithm and heuristic admission control scheme in less than 1.5 s of CPU time using distributed computation.<<ETX>>

Efficient Estimation and Collision-Group-Based Anticollision Algorithms for Dynamic Frame-Slotted ALOHA in RFID Networks

There are two challenges for the frame-slotted ALOHA algorithms in radio-frequency identification (RFID). The first challenge is estimating unknown tag-set size accurately; the second challenge is improving the efficiency of the arbitration process so that it uses less time slots to read all tags. This study proposes estimation algorithm based on the Poisson distribution theory and identifies the overestimation phenomenon in full collision. Our novel anticollision algorithm alternates two distinct reading cycles for dividing and solving tags in collision groups. This makes it more efficient for a reader to identify all tags within a small number of time slots.

Minimax end-to-end delay routing and capacity assignment for virtual circuit networks

We consider the routing and capacity assignment problem for virtual circuit networks where the objective is to minimize the maximum end-to-end packet delay. Compared with traditional aggregate type of performance measures, such as the average packet delay, this performance measure is consistent with those of many new services and achieves better fairness among users. Two cases are considered. In the first case, the capacity assignment is assumed to be given and the routing strategy is to be determined. The problem is formulated as a nonlinear nonconvex multicommodity network flow problem with integer routing decision variables. A dual approach is proposed to calculate primal feasible solutions. The proposed algorithm is computationally shown to uniformly outperform a greedy heuristic and a linear programming relaxation approach in conjunction with a rounding scheme. The algorithm also calculates legitimate lower bounds on the optimal objective function value, which are not easily attainable using the linear programming relaxation approach due to the nonconvex nature of the problem. In the second case, we consider the joint routing and capacity assignment problem. The problem is formulated as a nonlinear nonconvex mixed integer programming problem. We develop a two-phase algorithm where the routing and the capacity assignment decision variables are optimized, respectively. A minimum-hop heuristic is used to calculate the routing assignment, and then a convex programming procedure is devised to solve the capacity assignment problem. Experimental results of this heuristic and comparison with the results of the first case are presented.

A routing algorithm for virtual circuit data networks with multiple sessions per O - D pair

In virtual circuit networks, all the packets in a session are transmitted over exactly one path established between the origin and the destination. For each origin–destination pair, it is assumed that there are multiple sessions. We consider the problem of choosing a path for each session so as to minimize the average packet delay in the network. We formulate this problem as a nonlinear multicommodity flow problem with integer decision variables. An iterative scheme that is similar to local search is developed to solve this problem. In each iteration, we apply Lagrangean relaxation and a multiplier adjustment procedure to solve a restricted problem. We show that the Lagrangean dual problem can be solved exactly by solving a convex program. In computational experiments, our algorithm determines solutions that are within 1% of an optimal solution in minutes of CPU time for networks with 26–61 nodes. In addition, we show that our proposed algorithm is better both theoretically and computationally than K(0)-ordering, single-path routing, or round-off Frank–Wolfe heuristics.