A single TCP connection with a time varying capacity bottleneck link is considered. The bottleneck capacity varies arbitrarily and independently of the TCP connection traffic; it aims at reproducing typical situations where capacity is modulated by exogenous process, such as wireless channel with link adaptation or bursty channels. The key point is to understand the interplay of the TCP congestion control evolution and the time constants of the bottleneck link capacity time variation. Fluid modeling is used to describe the time evolution of the congestion window size and of the bottleneck buffer content with a completely general capacity time function, ns-2 based simulations are used as mean to assess the fluid model accuracy. Numerical results show the existence of a "resonance" phenomena, that can significantly degrade the TCP long term performance (even more than halved). The degradation depends on the ratio between the fundamental time constant of the link capacity variation and the TCP average round trip time.

In recent years, issues regarding the behavior of TCP in high-speed and long-distance networks have been extensively addressed in the networking research community, both because TCP is the most widespread transport protocol in the current Internet and because bandwidth-delay product continues to grow. The well known problem of TCP in high bandwidthdelay product networks is that the TCP Additive Increase Multiplicative Decrease AIMD probing mechanism is too slow in adapting the sending rate to the end-to-end available bandwidth. To overcome this problem, many modifications have been proposed such as FAST TCP [1], STCP [2], HSTCP [3], HTCP [4], BIC TCP [5] and CUBIC TCP [6]. The goal of this work is to investigate, by using both analytical models and simulation results, optimal sizing of the retransmission and out-of-order TCP buffers in case of modified TCP congestion control settings and to highlight differences between NewReno TCP and SACK TCP packet loss recovery schemes in terms of TCP internal buffers requirements. An important result is that the SACK option turns out to be particularly effective in reducing buffer requirements in the case of very high bandwidthdelay product links.

In this paper, we investigate interactions between TCP and wireless hybrid FEC/ARQ schemes. The aim is to understand what is the best configuration of the wireless link protocol in order to guarantee TCP performance and channel efficiency. Interactions between TCP and different link layer mechanisms are evaluated by means of an analytic model that reproduces: 1) a Rayleigh fading channel with FEC coding, 2) a generic selective repeat ARQ Protocol, and 3) the TCP behavior in a wired-cum-wireless network scenario. The analytic model is validated-by means of ns-based simulations. The analysis represents a contribution to the optimal design of link layer parameters of wireless networks crossed by TCP/IP traffic. The main findings can be summarized as follows: 1) fully reliable ARQ protocols are the best choice for both TCP performance and wireless link efficiency and 2) optimal values of FEC redundancy degree from the point of view of energy efficiency maximizes TCP performance as well.

In this work we address the problem of inferring the presence of a bottleneck from passive measurement in the UMTS core network. The study is based on one month of packet traces collected in the core network of mobilkom austria AG & Co KG, the leading mobile telecommunications provider in Austria, EU. During the measurement period a bottleneck link in the UMTS core network was revealed and removed, therefore the traces enable the accurate analysis and comparison of the traffic behavior in the two network conditions. The proposed approach exploits statistics of estimated TCP performance parameters (e.g. RTT, re-transmissions) in order to build a set of bottleneck indicators. We show that such statistics are volatile due to the presence of few top users, but this effect can be counteracted with a simple filtering method. Results show that the frequency of re-transmission events is a powerful indicator for the specific type of bottleneck under study, and it can be used to provide early warning about future occurrences of similar events. This application is particularly important for operational UMTS networks nowadays, since the traffic volumes and composition is still under evolution.

In this paper we investigate the interaction between TCP and wireless ARQ mechanisms. The aim is to understand what is the best reliability degree of the wireless link in order to guarantee TCP performance. For this purpose, we first develop a Markov model for a selective repeat ARQ protocol, widely used in the current wireless environments. Secondly, we design a cross-layer algorithm that, by exploiting the proposed model, can adapt the number of link layer transmission attempts to the end-to-end packet loss rate perceived by TCP. The interaction between TCP and link layer is evaluated in a specific case study (TCP over 3G radio access) by means of simulations carried out by using a very detailed UMTS-TDD simulator based on us. The deployment of the link layer Markov model and of the proposed algorithm allows us to derive some interesting conclusions about the design of retransmission protocols in TCP/IP network environments.

Wireless data is deemed as a major booster of next generation wireless networks. In this context, the support of Internet applications based on the TCP still presents several open issues. The aim of this work is to evaluate the performance of TCP data transfer over the UMTS TD-CDMA air interface, by means of a rather detailed simulation model. The evaluation includes both bulk FTP like data transfer and interactive client-server (WWW) traffic. The distinctive point of view is exploring the interaction between application and transport level protocol functions and lower layers protocols over the radio interface. The interplay of radio access and fixed core network impairments is analyzed as well. Major results deal with balancing the error recovery performance of the RLC against the radio channel correlation and the fixed network impairments.