07-06-2013, 12:15 PM
Congestion Minimization
Congestion.doc (Size: 29.5 KB / Downloads: 15)
Minimization.doc (Size: 23.5 KB / Downloads: 15)
Flow diagram.doc (Size: 25.5 KB / Downloads: 19)
Modules.doc (Size: 23 KB / Downloads: 15)
Project Description.doc (Size: 21.5 KB / Downloads: 16)
System Testing1.doc (Size: 26.5 KB / Downloads: 17)
Abstract:
Recent research efforts to design better Internet transport protocols combined with scalable Active Queue Management (AQM) have led to significant advances in congestion control. One of the hottest topics in this area is the design of discrete congestion control algorithms that are asymptotically stable under heterogeneous feedback delay and whose control equations do not explicitly depend on the RTTs of end-flows. In this paper, we first prove that single-link congestion control methods with a stable radial Jacobian remain stable under arbitrary feedback delay (including heterogeneous directional delays) and that the stability condition of such methods does not involve any of the delays. We then extend this result to generic networks with fixed consistent bottleneck assignments and max–min network feedback. To demonstrate the practicality of the obtained result, we change the original controller in Kelly et al.’s work to become robust under random feedback delay and fixed constants of the control equation. We call the resulting framework Max–min Kelly Control (MKC) and show that it offers smooth sending rate, exponential convergence to efficiency, and fast convergence to fairness, all of which make it appealing for future high-speed networks.
Existing System:
Existing system model is in a decentralized network, which describes two algorithms (primal and dual) and proves their global stability in the absence of feedback delay.
Proposed system:
Proposed system is the Internet congestion controls under non-negligible directional feedback delays.
We focused on the control methods with radial Jacobians, it shows that all such systems are stable under heterogeneous delays.
To construct a practical congestion control system with a radial (symmetric in particular) Jacobian, we made two changes to the classic discrete Kelly control and created a max–min version we call MKC.
Modules:
1. Packet header
2. The Router
3. Classic Kelly Control
4. Max–Min Kelly Control
5. Delay-Independent Stability
6. Single and Multi-Link Stability
1. Packet header:
The MKC packet header consists of two parts a 16-byte router header and a 4-byte user header. MKC router header encapsulates information that is necessary for the router to generate precise AQM feedback and subsequently for the end user to adjust its sending rate. The field is a unique label that identifies the router that generated the feedback (e.g., its IP address). This field is used by the flows to detect changes in bottlenecks, in which case they wait for an extra RTT before responding to congestion signals of the new router. The field is a local variable incremented by the router each time it produces a new value of packet loss (see below for more). Finally, the field carries the length of the averaging interval used by the router in its computation of feedback.
2. The Router:
The major task of the router is to generate its AQM feedback and insert it in the headers of all passing packets. However, notice that the router never knows the exact combined rate of incoming flows.
3. Classic Kelly Control
A single-source, single-link configuration and utilize a congestion indication function that computes the estimated packet loss using instantaneous arrival rates. The convergence speed of Kelly control is improved from linear to exponential. an important difference between real network paths, which are limited by the slowest resource, and the model of proportional fairness augmented , which takes into account the capacity of all resources in the network. This difference leads to significant overflow of slow routers and underutilization of fast routers along a given path.
4. Max–Min Kelly Control
To avoid unfairness between flows, one must fix the control parameters of all end users and establish a uniform set of equations that govern the system. Max–min Kelly Control (MKC) and demonstrate that its stability and fairness do not depend on any parameters of the network (such as delay, path length, or the routing matrix of end users).
5. Delay-Independent Stability
Before restricting our analysis to MKC, we examine a wide class of delayed control systems, whose stability directly follows from that of the corresponding un delayed systems. We subsequently show that MKC belongs to this category and obtain a very simple proof of its stability.
6. Single and Multi-Link Stability
The stationary point and has the same first-order partial derivative for all end users. Our goal is to derive sufficient and necessary conditions. This implementation
removes the oscillations that originally occurred and there is no oscillation in both the
transient phase and the steady state.