Design and development of dynamic queue storage and floating IP active measurement in IP networks

The Next Generation Network (NGN) has been anticipated to be an infrastructure of many newly emerging applications and majority of traditional services like legacy telephone networks that have been reshaped into the Internet services. Although NGN does not promise fundamental changes, the infrastructure is proposed in two main directions: wired and wireless networks, both relying on IPv6 as the network layer architecture. However, IPv6 still relies on TCP/IP architecture which has advanced with countless patches and temporary solutions. Although it is expected that in future Internet In this thesis an approach that utilizes routing loops and network delay to keep IP packets in the network in the form of Floating IP Packets (FIPs) is proposed. The development and validation of the proposed technique is given using the OpNet Modeller simulation tool and actual IPv4 and IPv6 test-beds. The two problems mentioned above will then be addressed based on the proposed FIP technique. As for the packet drop problem, a FIP-based Dynamic Queue Storage (DQS) for delay-tolerant traffic types is proposed. Applying DQS, the network can play the role of temporary non-physical data storage. Loop-delay dynamic storage capacity is discussed based on analytical model and simulation experiment and it is shown that DQS capacity only depends on the total delay and the maximum available andwidth/processing power of the routing loop. To address the measurement problem based on FIP, a concise measurement technique is proposed which is called Floating IP Packet Active Measurement (FIPAM). This technique utilizes routing loop for keeping a single packet the packet loss/drop due to network errors reduces, the intentional packet drop that occurs in non-real-time and delay tolerant traffic types increases significantly especially over the Access Networks. This is either due to extensive Quality of Service (QoS) and traffic/flow management schemes developed for real-time traffic or due to temporary out-of-coverage in wireless networks. Another problem is that it is essential for service providers to have a precise insight about the delivery path on the Internet. However, most of the current measurement techniques rely on ICMP protocol and terminal hosts. Majority of today’s network operators block ICMP traffic or give them least possible priority which leads to inaccurate or unsuccessful results. Furthermore, in future high speed networks, where the network speed exceeds the host’s I/O speed, host-based approaches significantly reduce the accuracy. In this thesis an approach that utilizes routing loops and network delay to keep IP packets in the network in the form of Floating IP Packets (FIPs) is proposed. The development and validation of the proposed technique is given using the OpNet Modeller simulation tool and actual IPv4 and IPv6 test-beds. The two problems mentioned above will then be addressed based on the proposed FIP technique. As for the packet drop problem, a FIP-based Dynamic Queue Storage (DQS) for delay-tolerant traffic types is proposed. Applying DQS, the network can play the role of temporary non-physical data storage. Loop-delay dynamic storage capacity is discussed based on analytical model and simulation experiment and it is shown that DQS capacity only depends on the total delay and the maximum available bandwidth/processing power of the routing loop. To address the measurement problem based on FIP, a concise measurement technique is proposed which is called Floating IP Packet Active Measurement (FIPAM). This technique utilizes routing loop for keeping a single packet in the network to calculate delay and throughput in the network. FIPAM uses the IP protocol and allows flexible packet size, packet format and QoS parameters. It is shown that the proposed DQS reduces packet drop arising from insufficient queue space by up to 6% with respect to loop parameters, without consuming more space in the physical queue from real-time traffic. However, it is also shown that by applying DQS, the total delay of the otherwise discarded packets increases with the maximum of 200 milliseconds. The results of keeping a single IPv4 and IPv6 packet on the test-bed network for more than 60 seconds are presented and it is shown by simulation that maximum 5,500 packets are kept on the network in a routing loop for more than 80 seconds without using physical queue. Results from comparing the proposed FIPAM with popular and widely used IPv4 Ping on an actual IPv4 test-bed show that FIPAM offers similar behaviour to Ping with up to 60% higher efficiency in measuring Round Trip Time (RTT) on the test-bed’s wired Ethernet LAN. Furthermore, several simulation scenarios similar to the test-bed have been developed in OpNet Modeller on several popular links. It is shown that the variation of the trend and values of the measured RTT based on FIPAM follows the same trend and values of the links’ speed.

Preview PDF FK 2011 29R.pdf Download (797kB) | Preview

Actions (login required)

View Item