Network Device Packet Loss Reduction During TCP Slow Start
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
The unpredictable nature of wireless connections often leads to packet loss during the slow start stage of a TCP session, disrupting network performance.
Innovation Solution
A network device transmits original and duplicate encapsulating packets through multiple connections using selection policies to maximize the chances of successful data transmission during the slow start stage, reducing packet loss and extending the duration of the exponential increase in the congestion window.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If duplicate packets are transmitted through multiple connections during slow start stage, then packet loss is reduced and transmission reliability is improved, but network device complexity and processing overhead increase
Solution Approach 1:
The patent applies preliminary action by proactively transmitting duplicate packets during the slow start stage before congestion can occur. The network device predicts potential packet loss in the vulnerable slow start phase and preemptively sends redundant copies through multiple connections, preventing the need for retransmission and maintaining reliability without requiring complex real-time congestion response mechanisms.
Solution Approach 2:
The patent segments the packet transmission process by dividing data packets into original encapsulating packets (OEPs) and duplicate encapsulating packets (DEPs). These segments are transmitted through different connections using different selection policies, allowing the system to distribute the reliability burden across multiple transmission paths rather than relying on a single complex error correction mechanism.
2Reliability
If multiple connections are used to transmit duplicate packets, then transmission reliability improves, but network bandwidth consumption increases
Solution Approach 1:
The patent applies partial action by transmitting duplicate packets only during the specific slow start stage rather than continuously. The duplication is temporary and targeted, occurring only when the congestion window is exponentially increasing and packet loss risk is highest. Once the slow start stage ends and steady-state transmission begins, duplicate transmission ceases, optimizing the trade-off between reliability and bandwidth consumption.
Solution Approach 2:
The patent changes the transmission parameters dynamically by adjusting the selection policy based on the TCP session stage. During slow start, multiple connections with duplicate packets are used; during steady state, single connection with original packets is used. This parameter change allows the system to optimize bandwidth consumption while maintaining reliability when needed.
3Duration of action of moving object
If duplicate encapsulating packets are transmitted through multiple connections, then the duration of slow start stage is extended, but network device processing time and complexity increase
Solution Approach 1:
The patent uses copying by creating duplicate encapsulating packets (DEPs) that are identical copies of the original data packets. These copies are transmitted through alternative connections if the original transmission is lost or delayed. The copying mechanism extends the slow start duration by ensuring packet delivery without requiring complex error correction or retransmission protocols, thereby reducing processing overhead despite the extended time.
4Productivity
If selection policies are applied for choosing connections, then transmission optimization is achieved, but control complexity and computational overhead increase
Solution Approach 1:
The patent applies preliminary action by pre-establishing multiple connections and defining selection policies before data transmission begins. The network device prepares the transmission path hierarchy in advance, so during actual transmission, simple policy-based routing is used rather than complex real-time optimization. This preliminary preparation reduces computational overhead while maintaining transmission efficiency.
Solution Approach 2:
The patent changes control parameters based on transmission stage by using different selection policies for OEPs and DEPs. The first selection policy for original packets and second selection policy for duplicate packets are simple, predefined rules that change based on packet type rather than complex dynamic optimization. This parameter-based control achieves transmission optimization with minimal computational overhead.
Data Source
AI summary
The present invention discloses a system and method for transmitting data packets at a network device. The network device of the system may select a first connection according to a first selection policy and select at least one second connection according to the second selection policy respectively if a data packet belonging to a new session is received. After the selection, network device may transmit each of original encapsulating packets (OEPs) through the first connection, and transmit each of at least one duplicate encapsulating packets (DEPs) through the at least one second connection within a specific period of time during a slow start stage. After the slow start stage, network device may transmit OEP only through a third connection, and no longer transmit the at least one DEP.