ICN: A New Approach to Internet Communication

Study Notes

TCP/IP-based Internet is designed for end-to-end communication, using location-based identifiers like IP addresses.
Internet traffic is experiencing a massive increase in data, with high levels of data replication.
The shift from wired to wireless networks necessitates better mobile communication support.
Existing IP address-based systems struggle with location changes, leading to increased overhead for mobility management.

Aims to address the limitations of IP by providing data more efficiently, supporting seamless mobility, and improving data security.
ICN transitions the communication model from request-to-location to request-for-content.
It views the Internet as a vast database and offers a new way of accessing data.

Employs a receiver-driven communication model with queries (interests) and data (responses).
Uses content identifiers rather than location identifiers.
Enables in-network caching at network nodes, leveraging routers for caching and other functions.
ICN's design allows for efficient content naming and retrieval.

Two major ICN architectures: Centralized and Decentralized.
Centralized:
- Utilizes a centralized Name Resolution System (NRS) for resolving names and publishing content.
- Uses a Flat Name Structure with caching at edge routers.
Decentralized:
- Includes Named Data Networking (NDN) and Content Centric Networking (CCN).

Content is addressed using a hierarchical name structure that resembles file system paths.

Messages:
- Interest (Query) packets contain information about the requested content, including unique identifiers, security details, and routing hints.
- Response (Data) packets contain the requested content, metadata, and security signatures.
Data Structures:
- Forwarding Information Base (FIB) – similar to routing tables, it maps name prefixes to next hops.
- Pending Interest Table (PIT) – stores pending queries for efficient response routing.
- Content Store (CS) – acts as a cache for storing content.

NDN's forwarding process involves matching interest names against entries in the FIB to determine the next hop.
NDN can utilize the same routing algorithms as IP but requires different protocols for managing Interest and Data messages.
NDN routing tables have a potential for larger size compared to IP's due to the ability to store individual content names.

IP Routing Table:
- Limits entries by aggregating individual addresses into prefixes for scalability.
- Faces challenges with growing table size due to factors like multi-homing, mobility, and traffic engineering.
NDN Routing Table:
- Also relies on name prefixes to manage entries efficiently.
- Leverages "map and encap" techniques to map application names to routable prefixes, typically belonging to ISPs.

IP Routing:
- Essential for packet delivery, requiring perfect routing paths and fast convergence.
NDN Routing:
- Acts as a supporting mechanism for content forwarding.
- Does not need to be perfect and supports a more flexible routing approach.

IP:
- Lacks explicit loop detection mechanisms.
- Relies on TTL (Time to Live) to prevent loops.
NDN:
- Employs PIT to track Interests and utilize unique nonces (random numbers) in each Interest packet to identify loops.

IP:
- Uses one-way traffic and a stateless approach, limiting failure detection capabilities.
NDN:
- Tracks response times and PIT entries to identify potential failures.
- Supports explicit Interest NACK to signal problems upstream.

Multipath Forwarding:
- Easier in NDN due to its two-way traffic and datagram state.
Adaptive Forwarding:
- Enabled by named data, two-way traffic, and datagram state.
- Efficiently handles link failures, congestion, and hijacking attempts.

IP:
- Location-bound address system limits mobility support.
- Faces challenges with latency, unreliability, and cost associated with mobility management.
NDN:
- Content-based addressing inherently supports customer mobility.
- Producer mobility can be addressed through rendezvous points.

Content-centric addressing: simplifies content naming and retrieval.
In-network caching: enhances content availability and reduces network load.
Mobility support: enables seamless mobility for both users and producers.
Security enhancements: offers robust security features with content signatures and encryption.