Label Distribution Protocol

Abstract: This document describes the use of RSVP (Resource Reservation Protocol), including all the necessary extensions, to establish label-switched paths (LSPs) in MPLS (Multi-Protocol Label Switching) Since the flow along an LSP is completely identified by the label applied at the ingress node of the path, these paths may be treated as tunnels A key application of LSP tunnels is traffic engineering with MPLS as specified in RFC 2702

Abstract: Generalized multiprotocol label switching, also referred to as multiprotocol lambda switching, supports not only devices that perform packet switching, but also those that perform switching in the time, wavelength, and space domains. The development of GMPLS requires modifications to current signaling and routing protocols. It has also triggered the development of new protocols such as the Link Management protocol. We present the traffic engineering enhancements to the Open Shortest Path First Internet routing protocol and ISIS Intradomain Routing Protocol, two popular routing protocols, to support GMPLS. We present the concepts of generalized interfaces, label-switched path hierarchy, and link bundling intended to improve GMPLS scalability. We also discuss the Link Management Protocol which can be used to make the underlying links more manageable.

Abstract: Generalized multiprotocol label switching (GMPLS), also referred to as multiprotocol lambda switching, is a multipurpose control plane paradigm that supports not only devices that perform packet switching, but also devices that perform switching in the time, wavelength, and space domains. The development of GMPLS necessitates enhancements to existing IP signaling and routing protocols. We present enhancements to two commonly used IP signaling protocols, RSVP and LDP, to support GMPLS. We illustrate the concept of hierarchical label switched path setup with an example, discuss mechanisms for bidirectional LSP setup, and describe the applications of suggested labels. We also discuss the important problem of protection and restoration in the GMPLS context. Finally, we describe how various recovery mechanisms can be implemented within the GMPLS framework.

Abstract: This paper presents LDP-Fed, a novel federated learning system with a formal privacy guarantee using local differential privacy (LDP). Existing LDP protocols are developed primarily to ensure data privacy in the collection of single numerical or categorical values, such as click count in Web access logs. However, in federated learning model parameter updates are collected iteratively from each participant and consist of high dimensional, continuous values with high precision (10s of digits after the decimal point), making existing LDP protocols inapplicable. To address this challenge in LDP-Fed, we design and develop two novel approaches. First, LDP-Fed's LDP Module provides a formal differential privacy guarantee for the repeated collection of model training parameters in the federated training of large-scale neural networks over multiple individual participants' private datasets. Second, LDP-Fed implements a suite of selection and filtering techniques for perturbing and sharing select parameter updates with the parameter server. We validate our system deployed with a condensed LDP protocol in training deep neural networks on public data. We compare this version of LDP-Fed, coined CLDP-Fed, with other state-of-the-art approaches with respect to model accuracy, privacy preservation, and system capabilities.