Patent No. US10708168 (titled "Routing methods, systems, and computer program products") on Mar 20, 2020. The application was issued on Jul 7, 2020.
’168 is related to the field of network routing and packet forwarding, specifically within the context of segment routing (SR) architectures. It addresses the challenges of managing traffic flow in increasingly large and complex networks where traditional destination-based routing often lacks the granularity required for precise path control. The background context involves the evolution of the Internet Protocol and the need for more flexible mechanisms to direct data through specific sequences of nodes or links to optimize latency and bandwidth.
The underlying idea behind ’168 is the transformation of a high-level explicit path definition into a machine-executable list of segment identifiers that can be embedded directly into a packet header. Rather than relying on each intermediate router to make an independent forwarding decision based solely on a final destination, the invention allows a source node to pre-calculate and encode the entire journey. This insight shifts the intelligence to the network edge, enabling source-directed routing that is both scalable and compatible with existing forwarding table structures.
The claims of ’168 focus on a method and system for generating a set of segment identifiers that uniquely encode an explicit path between two nodes. The independent claims specify that this generation process involves selecting between two distinct types of identifiers: a nodal segment identifier, which uniquely identifies a specific node in the network, and an adjacency segment identifier, which identifies a specific link between two neighboring nodes. These identifiers are then stored at the ingress node and inserted into the packet header to dictate the packet's traversal.
In practice, the system operates by receiving a sequence of nodes or links that define a desired route. The ingress node then constructs a segment list by mapping this path to the corresponding identifiers found in the network's segment routing forwarding tables. When a packet is ready for transmission, the node prepends this list to the header. As the packet moves through the network, each intermediate node simply looks at the top identifier in the list to determine the next hop, effectively following the pre-baked instructions without needing to calculate the path itself.
This approach differs from prior solutions by providing a hybrid identification scheme that can toggle between node-based and link-based routing within the same path encoding. Unlike traditional source routing, which often requires large headers containing full IP addresses, this invention uses compact segment identifiers that are indexed in local forwarding tables. This reduces header overhead and allows for more granular traffic engineering, such as forcing a packet over a specific physical link (via adjacency SIDs) or simply directing it to a specific processing node (via nodal SIDs).
In the early 2010s when ’168 was filed, network architectures were typically implemented using a rigid distinction between names, addresses, and routes, where protocol addresses primarily served as location identifiers rather than path descriptors. At a time when systems commonly relied on global 32-bit or 128-bit address spaces to reach endpoints, the mapping of these addresses to specific physical routes was generally relegated to lower-level gateway procedures and autonomous routing tables. During this era, hardware and software constraints made the dynamic encoding of explicit network paths directly into protocol addresses non-trivial, as standard routing logic was optimized for hop-by-hop forwarding based on fixed-length destination identifiers rather than source-defined path information.
The disclosed invention represents a meaningful technical advancement through an architectural shift that integrates path information directly into the protocol address structure. By generating a set of segment identifiers that encode an explicit sequence of nodes or links, the system enables a source node to define a specific traversal path for a packet rather than relying on independent hop-by-hop decisions by intermediate routers. This integration is achieved by selecting between nodal segment identifiers, which uniquely identify single nodes, and adjacency segment identifiers, which identify specific links between contiguous nodes. The resulting technical effect is the ability to store and insert these path-encoded segment lists into packet headers, overcoming the constraints of traditional address-route separation and providing precise control over network traffic engineering and path selection.
This patent contains 26 claims, with claims 1 and 26 serving as the independent claims. The independent claims focus on a method and a system for encoding an explicit network path into a set of segment identifiers, which are selected from nodal or adjacency types and added to a packet header to guide it through a sequence of nodes or links. The dependent claims generally serve to refine the path encoding process by specifying shortest path calculations, defining the use of nodal and adjacency segments to minimize the number of identifiers, and detailing the configuration of identifier spaces and forwarding tables within the segment routing framework.
Definitions of key terms used in the patent claims.
US Latest litigation cases involving this patent.

The dossier documents provide a comprehensive record of the patent's prosecution history - including filings, correspondence, and decisions made by patent offices - and are crucial for understanding the patent's legal journey and any challenges it may have faced during examination.
Get instant alerts for new documents