Search

In-camera event denoising reduces the data rate of event cameras by filtering out noise at the source. A lightweight multilayer perceptron denoising filter (MLPF) provides state-of-the-art low-cost denoising accuracy. It processes a small neighborhood of pixels from the timestamp image around each event to discriminate signal and noise events. This paper proposes two digital logic implementations of the MLPF denoiser and quantifies their resource cost, power, and latency. The hardware MLPF quantizes the weights and hidden unit activations to 4 bits and has about 1k weights with about 40% sparsity. The Area-Under-Curve Receiver Operating Characteristic accuracy is nearly indistinguishable from that of the floating point network. The FPGA MLPF processes each event in 10 clock cycles. In FPGA, it uses 3.5k flip flops and 11.5k LUTs. Our ASIC implementation in 65nm digital technology for a 346x260 pixel camera occupies an area of 4.3mm^2 and consumes 4nJ of energy per event at event rates up to 25MHz. The MLPF can be easily integrated into an event camera using an FPGA or as an ASIC directly on the camera chip or in the same package. This denoising could dramatically reduce the energy consumed by the communication and host processor and open new areas of always-on event camera application under scavenged and battery power. Code: https://github.com/SensorsINI/dnd_hls, Comment: Accepted to 2023 CVPRW Workshop on Event-Based Vision

Peer-to-peer network is one in which each node in the network can act as a client or server for the other nodes in the network. It allows shared access to various resources such as files, peripherals, and sensors without the need for a central server. Content distribution in the P2P network from server is done by multicasting. Multicasting is the process of sending the data to the multiple designations. This technology is highly efficient for the large scale multimedia content delivery in P2P network where the end peer have identical set of system components. But in reality, the peers have heterogeneous set of requirements for different service levels as well as different service components. The ability to provide differentiated services to each peer with widely varying requirements is becoming important. We need to provide differentiated Services above the existing shared network infrastructure. The solution proposed to solve the above said problem is to provide individualized service to each peer. It focuses on constructing and maintaining an efficient multiple overlay multicast tree structure in the P2P network. The tree maintenance process is governed by two mechanisms called as dynamic reconfiguration driven by peer and less frequent tree maintenance by network status change observation. In this paper new scalable architecture is constructed and analysed based on the above strategies., Comment: 6 pages,5 figures

Most of the existing P2P content distribution schemes implement a random or rarest piece first dissemination procedure to avoid duplicate transmission of the same pieces of data and rare pieces of data occurring in the network. This problem can be solved using peer-to-peer content distribution based on network coding scheme. Network coding scheme uses random linear combination of coded pieces. Hence the above stated problem can be solved ease and simple. Our proposed mechanism uses network coding mechanism in which several contents of same message is grouped into different group and coding operation is performed only within the same group. The interested peers are also divided into several groups with each group have the responsibility to spread one set of contents of some message. The coding system is designed to assure the property that any subset of the messages can be utilized to decode the original content as long as the size of the subset is suitably large. To meet this condition, dynamic smart network coding scheme is defined which assures the preferred property, then peers are connected in the same group to send the corresponding message, and connect peers in different groups to disseminate messages for carrying out decoding operation. Moreover, the proposed system can be readily expanded to support topology change to get better system performance further in terms of reliability, link stress and throughput. The simulation results prove that the proposed system can attain 20 to 25% higher throughput than existing systems, it also further attains good reliability, link failure and robustness to peer churn., Comment: this article consists of 14 pages

Using the approach of autonomous Hamiltonian dynamical system, we attempt to estimate the as yet unknown upper limit on the size of the galactic halo surrounding galaxies (lenses). The key to determine the size of the halo is to determine the maximum radius up to which stable material circular orbits are admissible. We shall illustrate the approach by considering a solution of the Weyl gravity containing a halo parameter $\gamma$. The upper limit for several observed lenses are calculated for a typical value of $gamma$ for definiteness, with and without the cosmological constant $\Lambda $. These lenses (all having Einstein radius $R_{\text{E}}\approx 10^{23}$ cm) consistently yield an upper limit $R_{\text{max}}^{\text{stable}}$($\simeq 4.25\times 10^{27}$ cm) inside the de Sitter radius only when $\gamma$ is \textit{negative}, while a positive $\gamma$ yields $R_{\text{max}}^{\text{% stable}}$ always exceeding the de Sitter radius., Comment: 7 pages

Glasses are structurally liquid-like, but mechanically solid-like. Most attempts to understand glasses start from liquid state theory. Here we take the opposite point of view, and use concepts from solid state physics. We determine the vibrational modes of a colloidal glass experimentally, and find soft low-frequency modes that are very different in nature from the usual acoustic vibrations of ordinary solids. These modes extend over surprisingly large length scales.

From event-driven simulations of a gravity-driven channel flow of inelastic hard-disks, we show that the velocity distribution function remains close to a Gaussian for a wide range densities (even when the Knudsen number is of order one) if the walls are smooth and the particle collisions are nearly elastic. For dense flows, a transition from a Gaussian to a power-law distribution for the high velocity tails occurs with increasing dissipation in the center of the channel, irrespective of wall-roughness. For a rough wall, the near-wall distribution functions are distinctly different from those in the bulk even in the quasielastic limit.