Abstract—Weak state routing using decay Bloom filters has been widely studied in the field of data-oriented networks. The existing weak state routing schemes cannot facilitate in-network queries effectively. Given a query for any item at an arbitrary node, the noise 1 in an unrelated routing entries is very likely equal to the useful information in the right routing entries. Consequently, queries are routed by means of network flooding, which differs a lot from the desired way of weak state routing, irrespective of the network topology and the usage and decay models of the Bloom filters. This work addresses the root cause of the mismatch between the practically reachable performance of the existing weak state routing schemes and the desired performance. Specifically, we study the impact of decay model on the membership information in the routing entries, and evaluate the negative impact of noise on a routing decision. Based on such analytical results, we derive the necessary and sufficient condition of a feasible weak state routing using decay Bloom filters. Accordingly, we design a novel receiver-oriented approach for Bloom filters, called Wader, which satisfies the above condition. The simulation results match well with our theoretical analysis, which demonstrate that Wader guarantees the correctness and efficiency of weak state routing with high probability. I.

In this paper, we present an exact active statistics counter architecture called BRICK (Bucketized Rank Indexed Counters) that can efficiently store per-flow variable-width statistics counters entirely in SRAM while supporting both fast updates and lookups (e.g., 40 Gb/s line rates). BRICK exploits statistical multiplexing by randomly bundling counters into small fixed-size buckets and supports dynamic sizing of counters by employing an innovative indexing scheme called rank-indexing. Experiments with Internet traces show that our solution can indeed maintain large arrays of exact active statistics counters with moderate amounts of SRAM.

In this paper, we present an exact active statistics counter architecture called BRICK (Bucketized Rank Indexed Counters) that can efficiently store per-flow variable-width statistics counters entirely in SRAM while supporting both fast updates and lookups (e.g., 40 Gb/s line rates). BRICK exploits statistical multiplexing by randomly bundling counters into small fixed-size buckets and supports dynamic sizing of counters by employing an innovative indexing scheme called rank-indexing. Experiments with Internet traces show that our solution can indeed maintain large arrays of exact active statistics counters with moderate amounts of SRAM.

In this paper, we present an exact active statistics counter architecture called BRICK (Bucketized Rank Indexed Counters) that can efficiently store per-flow variable-width statistics counters entirely in SRAM while supporting both fast updates and lookups (e.g., 40 Gb/s line rates). BRICK exploits statistical multiplexing by randomly bundling counters into small fixed-size buckets and supports dynamic sizing of counters by employing an innovative indexing scheme called rank-indexing. Experiments with Internet traces show that our solution can indeed maintain large arrays of exact active statistics counters with moderate amounts of SRAM.

In this paper, we present an exact active statistics counter architecture called BRICK (Bucketized Rank Indexed Counters) that can efficiently store per-flow variable-width statistics counters entirely in SRAM while supporting both fast updates and lookups (e.g., 40 Gb/s line rates). BRICK exploits statistical multiplexing by randomly bundling counters into small fixed-size buckets and supports dynamic sizing of counters by employing an innovative indexing scheme called rank-indexing. Experiments with Internet traces show that our solution can indeed maintain large arrays of exact active statistics counters with moderate amounts of SRAM.

Abstract not found

Abstract—Gradient-based routing using Bloom filters is an effective mechanism to enable data-driven queries in multi-hop networks. A node compressively describes its data items as a Bloom filter, which is then diffused away to the other nodes with information decay. The Bloom filters form an information potential that eventually navigates queries to the source node by ascending the potential field. The existing designs of Bloom filters, however, have critical limitations with respect to the feasibility of gradient-based routing. The compressed routing entries appear to be noisy. Noise in unrelated routing entries is very likely to equal to even outweigh information in right routing entries, thus blinding a query to its desired destination. This work addresses the root cause of the mismatch between the idea and the practical performance of gradient-based routing using Bloom filters. We first investigate the impact of decaying model on the effectiveness of routing entries, and then evaluate the negative impact of noise on routing decisions. Based on such analytical results, we derive the necessary and sufficient condition of feasible gradient-based routing using Bloom filters. Accordingly, we propose a receiveroriented design of Bloom filters, called Wader, which satisfies the above condition. The evaluation results demonstrate that Wader guarantees the correctness and efficiency of gradient-based routing with high probability. I.