A Strategy of Designing Routing Algorithms Based on Ideal Routings

Abstract

This paper proposes a strategy of designing routing algorithms for connectionless packet-switched networks. This strategy consists of three design elements as follows: [A] the notion of ideal routings is introduced to provide the upper performance limits attained by improving routing algorithm and it serves as a standard to measure the performance of other algorithms; [B] a method of constructing simple algorithms is presented under implementation conditions from ideal routings; [C] a method is described to enhance the performance limits of [A]. By using these elements, simple algorithms with a maximum degree of performance attainment are realized. By "degree of performance attainment", we mean that we can see how much room is left for the improvement of algorithms. We develop [A] and [B] with the performance measures of throughput and average packet delay and the M/M/1 queuing network. We decide ideal static routings and their performance limits from [A]. We obtain a new simple algorithm from [B] based on the notion of the ideal routings in implementation conditions. The designed algorithm improves the throughput and the average delay, which are comparable to those from ideal static routings. This improvement is contrasted to the adaptive and distributed OSPF (Open Shortest Path First), a standard Internet routing protocol.

Modeling communication dynamics

“…This approach is common in information theoretic analysis of communication systems [3]. Some work on routing has also been done from this point of view [4]. In order to illustrate analysis of bounds of behavior, we present some examples from recent practical problems in wireless communications.…”

Hamiltonian algorithm with momentum attenuation for adaptive beamforming of ESPAR antenna

“…For the given initial values of reactances x i and values of momenta p i (i = 1, 2, …, M), and the values of control parameters A 0 and J, we update the values of x i n+1 , p i n+1 by iterations from n=1 to N , according to (5) and (8). Finally, we obtain the optimal reactance vector x that gives the minimum of the cost function.…”

“…Hamiltonian algorithm (HA) [7] [8] intends to meet the two conflicting requirements, i.e., to be deterministic and to be free from local problems. Hamiltonian algorithm originally stems from a heuristic idea with autonomous motion of mass in a friction-free potential space.…”

Performance Evaluation for RF-Combining Diversity Antenna Configured with Variable Capacitors

“…In digital signal processing, the optimum solution of the MRC (Maximal Ratio Combining) criteria realizes the maximum output SNR. On the other hand, in the RF-DA, convergence to the global maximum is difficult to achieve unless we use some methods such as the simulated annealing [6] or the Hamiltonian algorithm [7], [8] and so on. In this paper, we focus on how fast convergence can be achieved and do not emphasize whether the solution gives a global maximum or local.…”