Emi Garcia-Palacios

Host-Based P2P Flow Identification and Use in Real-Time

Data identification and classification is a key task for any Internet Service Provider (ISP) or network administrator. As port fluctuation and encryption become more common in P2P applications wishing to avoid identification, new strategies must be developed to detect and classify their flows. This article introduces a method of separating P2P and standard web traffic that can be applied as part of an offline data analysis process, based on the activity of the hosts on the network. Heuristics are analyzed and a classification system proposed that focuses on classifying those “long” flows that transfer most of the bytes across a network. The accuracy of the system is then tested using real network traffic from a core Internet router showing misclassification rates as low as 0.54% of flows in some cases. We expand on this proposed strategy to investigate its relevance to real-time, early classification problems. New proposals are made and the results of real-time experiments are compared to those obtained in the offline analysis. It is shown that classification accuracies in the real-time strategy are similar to those achieved in offline analysis with a large portion of the total web and P2P flows correctly identified.

Characterization of inter-body interference in context aware body area networking (CABAN)

The characterization and understanding of body to body communication channels is a pivotal step in the development of emerging wireless applications such as ad-hoc personnel localisation and context aware body area networks (CABAN). The latter is a recent innovation where the inherent mobility of body area networks can be used to improve the coexistence of multiple co-located BAN users. Rather than simply accepting reductions in communication performance, sensed changes in inter-network co-channel interference levels may facilitate intelligent inter-networking; for example merging or splitting with other BANs that remain in the same domain. This paper investigates the inter-body interference using controlled measurements of the full mesh interconnectivity between two ambulatory BANs operating in the same environment at 2.45 GHz. Each of the twelve network nodes reported received signal strength to allow for the creation of carrier to interference ratio time series with an overall entire mesh sampling period of 54 ms. The results indicate that even with two mobile networks, it is possible to identify the onset of co-channel interference as the BAN users move towards each other and, similarly, the transition to more favourable physical layer channel conditions as they move apart.

Fading characterization for Context Aware Body Area Networks (CABAN) in interactive smart environments

Body Area Networks are unique in that the large-scale mobility of users allows the network itself to travel across a diverse range of operating domains. This presents the possibility of creating interactive smart environments where Context Aware Body Area Networks can sense and co-operate with nearby wireless networks. This paper describes an investigation of the physical layer characteristics for a user within a ‘smart’ office environment. A series of carefully controlled measurements of the mesh interconnectivity both within and between an ambulatory body area network and a stationary desk-based network were performed using 2.45 GHz nodes. The measurement system was capable of simultaneously characterising both off-body and on-body channels. The results show that model selection for fading characterization was often arbitrary due to the divergence of the empirical data to that of the fading distributions. It has also been identified that not all the reciprocal links follow the same distribution, a concern for physical layer protocol design.

Assessing capacity in WLAN-UMTS integrated networks

In the very near future wireless LAN and UMTS technologies will have to co-exist in many populated areas. Since UMTS radio cells are more limited in capacity than WLAN access points (AP), this technology integration gives operators the opportunity to perform internetwork load balancing by transferring services to WLAN hotspots. By doing so, UMTS cell congestion can be alleviated and capacity can be freed for future real time connections. Several scenarios arise depending on the level of congestion of both networks. This research studies such scenarios by assessing the impact of performing internetwork load balancing when transferring non-real time traffic. Promising results show how light and intermediate loaded neighboring WLAN hotspots can absorb up to 50% of the load of a congested UMTS cell, enhancing the throughput for switched non-real time applications and freeing 50% capacity of for future real time UMTS calls. This is achieved without degrading the service of existing applications in the WLAN hotspot.

Architecture and implementation of a novel tag computation circuit for broadband wireless access packet scheduling

In this paper we present the hardware architecture and implementation of a tag computation circuit for a credit based Self-Clocked Fair Queuing (SCFQ) scheduler specifically targeted for packet scheduling in Broadband Wireless Access (BWA) as described in the recently released IEEE 802.16 Standards. Our objective is the implementation of a configurable scheduler that is based on the principles of weighted fair queuing combined with a credit based bandwidth reallocation scheme. The implementation provides the hardware platform for a runtime configurable scheduling architecture that is able to reallocate bandwidth on the fly if particular links should suffer packet loss due to unexpected noise or channel quality degradation. The system is implemented using FPGA technology and provides extended programmability to adapt the tag computation to a range of custom scheduling schemes. The hardware architecture is parallel and pipelined enabling an aggregated throughput rate of 180 million tag computations per second. The throughput performance is ideal for BWA nodes, allowing room for relatively complex computations in QoS aware adaptive scheduling. The high-level system breakdown is described and synthesis results for Xilinx FPGA technology are presented.

Distributed resource management and admission control in wireless ad hoc networks: a practical approach

The authors propose a novel and practical approach to estimate resources and perform a distributed admission control in multi-hop ad hoc networks based on multi-rate enabled IEEE 802.11 technology. The main challenge is to determine if there exist sufficient resources [i.e. the available bandwidth (AB)] for a new incoming flow to be admitted rather than quantifying the exact amount of existing resources. In order to determine the AB along a multi-hop path, the authors take into consideration the channel rate at each hop as well as the channel idle ratio of relevant neighbouring nodes. Furthermore, the admission control is performed at the same time as the AB is determined which minimises overhead. The proposed approach can be applied hop-by-hop in a distributed manner by the end-user, thus being suitable for wireless ad hoc networks. Analysis and simulation based on the Network Simulator version 2 (NS2) platform verify the proposed approach.

Identification of P2P flows through host activity

With the increasing quantity and varying nature of traffic crossing the internet, coupled with techniques such as fluctuating port numbers and transport layer encryption, the identification of individual packet flows is becoming more difficult. We introduce and investigate a new method for the detection of P2P flows based on the activity of the hosts (IP addresses) involved in the connection. Heuristics are generated that examine properties of these hosts and used to uniquely detect individual P2P and non-P2P flows. The identification strategy has been tested on two real network data traces from a core internet router with some classification accuracies showing higher than 99%.

Decomposition of a WFQ tag computation architecture

The paper investigates a decomposition of the WFQ algorithm in order that a high throughput scheduler may be produced. This is intended to be the basis for a parallel finishing tag computation hardware using the latest silicon technology. Significant acceleration of the tag computation is expected due to the use of multiple lookup tables in the form of embedded memory with independent control circuitry.