Semi-Explicit Construction of Linear Size Concentrators and Superconcentrators
نویسندگان
چکیده
Concentration and superconcentration operations play a central role in the constructions of nonblocking switching networks. This was established in a seminal effort by Bassalygo and Pinsker who showed that an n-input nonblocking switching network can be constructed with O(n log n) crosspoints if concentrators can be constructed with O(n) crosspoints. Starting with the pioneering work of Margulis, concentrators and superconcentrators with O(n) crosspoints were reported in the literature, but the constants in the crosspoint complexities of these concentrators are considerably large. In this paper, we extend the probabilistic constructions of Pinsker, Bassalygo and Pippenger to obtain semi-explicit constructions of concentrators and superconcentrators with fewer crosspoints. More specifically, we give a semi-explicit version of Pinsker’s bounded capacity concentrator to show that there exists a concentrator with at most 25n crosspoints. This improves Pinsker’s original bound of 29n crosspoints. Next we optimize Bassalygo’s concentrator construction to prove that there exists a concentrator with 19n crosspoints, improving his bound of 20n crosspoints. We then give a semi-explicit (n, 3n/4, n/2)-concentrator cosntruction with 4.5n crosspoints, and use this bounded capacity concentrator to obtain an n-superconcentrator with 40n crosspoints. Finally, we show that Bassalygo’s superconcentrator construction can be optimized to obtain a superconcentrator with 33n crosspoints.
منابع مشابه
Eigenvalues, Expanders and Superconcentrators
Explicit construction of families of linear expanders and superconcentrators is relevant to theoretical computer science in several ways. There is essentially only one known explicit construction. Here we show a correspondence between the eigenvalues of the adjacency matrix of a graph and its expansion properties, and combine it with results on Group Representations to obtain many new examples ...
متن کاملHigh performance concentrators and superconcentrators using multiplexing schemes
Concentrators are used to interface and combine together low speed communication channels onto higher speed transmission links to alleviate transmission costs. They are also used to construct more powerful switching fabrics such as permutation and broadcast networks. Using an adaptive binary sorting network model, this paper constructs new concentrators and superconcentrators. Unlike the previo...
متن کاملSuperconcentrators
An n-superconcentrator is an acyclic directed graph with n inputs and n outputs for which, for every -<_ n, every set of inputs, and every set of outputs, there exists an r-flow (a set of vertex-disjoint directed paths) from the given inputs to the given outputs. We show that there exist n-superconcentrators with 39n + O(log n) (in fact, at most 40n) edges, depth O(log n), and maximum degree (i...
متن کاملBetter Expanders and Superconcentrators by Kolmogorov Complexity
We show the existence of various versions of expander graphs using Kolmogorov complexity. This method seems superior to the usual “probabilistic construction”. Also, the best known bounds on the size of expanders and superconcentrators can be obtained this way. In the case of (acyclic) superconcentrators we obtain the density 34. Also, we review related graph properties, like magnification, edg...
متن کاملConcentration properties of semi-vertex transitive graphs and random bi-coset graphs∗
It is well-known that concentrators are sparse graphs of high connectivity, which play a key role in the construction of switching networks; and any semi-vertex transitive graph is isomorphic to a bi-coset graph. In this paper, we prove that random bi-coset graphs are almost always concentrators, and construct some examples of semi-vertex transitive concentrators.
متن کامل