Jarmo Harju

Cross-layer modeling of wireless channels for data-link and IP layer performance evaluation

To provide a tool for performance evaluation of IP-based delay- and loss-sensitive applications running over wireless channels we propose a novel cross-layer wireless channel modeling approach. We firstly develop simple and computationally efficient wireless channel modeling algorithm. For this purpose we adopt the special solution of the inverse eigenvalue problem and show that its complexity significantly decreases when the time-series is covariance stationary two-valued in nature. Our model explicitly takes into account autocorrelation and distributional properties of empirical data. Then, we extend this model to IP layer using the cross-layer mappings. The resulting model is represented by the IP packet error process and reflects memory properties of initial bit error process. We show that our approach allows to get accurate estimators of IP packet error probabilities in presence of FEC at the data-link layer eliminating the need for computationally expensive time-consuming bit level simulations. It also provides a way to choose the required correction capabilities of FEC codes resulting in best possible performance at the data-link and IP layers.

Peer-to-Peer Streaming Technology Survey

Lately there has been a growing interest in the use of peer-to-peer technologies for deploying large-scale live media streaming systems over the Internet. In this paper we give a brief survey on the peer-to-peer streaming field and have also a closer look on selected applications. In practice, we analyse the selected peer-to-peer streaming systems (Octoshape, SopCast, TVAnts and TVU networks) and see if they are suitable for mobile usage. This paper also shows results from an experimental test carried out by using selected applications with a PC over different network connections (EDGE, UMTS, HSDPA, ADSL and LAN). None of the selected applications are designed to be used in a mobile environment, so there is still a lot of work to do to have optimized systems in the mobile network environment.

TCP NewReno Throughput in the Presence of Correlated Losses: The Slow-but-Steady Variant

This paper presents an analytical model of steady state throughput of the Slow-but-Steady variant of TCP NewReno as a function of loss event rate, average number of segments lost per loss event, average round trip time, and retransmission timeout value. The presented model is based on the approach proposed by Padhye et al. for TCP Reno throughput modeling and extends it by capturing the effect of fast recovery algorithm of the Slow-but-Steady variant of TCP NewReno on throughput and taking into consideration slow start phase after timeout expiration. Validation by ns-2 simulations shows good agreement and accuracy of the proposed model.

Co-operation and comparison of DiffServ and IntServ: performance measurements

IETF has developed two architectures for the Internet, Integrated Services (IntServ) and Differentiated Services (DiffServ), which should enable QoS based handling of data flows in IP networks. This paper presents a number of measurements which were carried out in order to determine some performance differences between these QoS mechanisms, as well as some characteristics about their joint use. The results show that in some cases the flow aggregation model, i.e. DiffServ, can provide the same or even better QoS than the reservation based model, i.e. IntServ.

Modeling TCP SACK performance over wireless channels with completely reliable ARQ/FEC

SUMMARY In this paper, we propose an analytical cross-layer model for a Transmission Control Protocol (TCP) connection running over a covariance-stationary wireless channel with a completely reliable Automatic Repeat reQuest scheme combined with Forward Error Correction (FEC) coding. Since backbone networks today are highly overprovisioned, we assume that the wireless channel is the only one bottleneck in the system which causes packets to be buffered at the wired/wireless interface and dropped as a result of buffer overflow. We develop the model in two steps. At the first step, we consider the service process of the wireless channel and derive the probability distribution of the time required to successfully transmit an IP packet over the wireless channel. This distribution is used at the next step of the modeling, where we derive expressions for the TCP long-term steady-state throughput, the mean round-trip time, and the spurious timeout probability. The developed model allows to quantify the joint effect of many implementationspecific parameters on the TCP performance over both correlated and non-correlated wireless channels. We also demonstrate that TCP spurious timeouts, reported in some empirical studies, do not occur when wireless channel conditions are covariance-stationary and their presence in those measurements should be attributed to non-stationary behavior of the wireless channel characteristics. Copyright 2011 John Wiley & Sons, Ltd.

Performance response of wireless channels for quantitatively different loss and arrival statistics

In this paper we propose a cross-layer performance evaluation framework for wireless channels and subsequently explore the performance response at the FEC and ARQ enabled data-link layer in terms of frame losses and delays for different first- and second-order error and arrival statistics. For a wireless channel model to be appropriate for various FEC capabilities without the need for extensive measurements of the frame error process for each particular FEC code, the error process of the wireless channel is modeled at the physical layer. We assume weak stationary property for bit error observations and model them using a hidden Markov model. The associated parameters matching algorithm allows us to explicitly capture bit error rate and lag-1 autocorrelation of the bit error process. To explore the performance response of the wireless channel at the data-link layer, a cross-layer extension of the bit error model to the data-link layer is then developed. The performance of applications is evaluated using the queuing-theoretic approach that allows both arrival and error processes to be autocorrelated and still retains analytical tractability. The proposed methodology allows us to obtain estimators of frame loss and delay probabilities in the presence of FEC and ARQ procedures at the data-link layer eliminating the need for time-consuming simulations and extensive measurements of wireless channel characteristics for different error correction capabilities of the data-link layer. It is analytical in nature, efficient for small and moderate frame sizes and suitable for performance control purposes where fixed size frames are used at the data-link layer. Particularly, it provides a way to choose the required correction capability of the FEC code resulting in best possible performance at the data-link layer. Numerical results indicate that first- and second-order bit error and frame arrival statistics significantly affect performance parameters provided by a wireless channel and should be taken into account when choosing an appropriate correction capability of the FEC code for given wireless channel conditions.

Factoring IPv6 Device Mobility and Ad-hoc Interactions into the Service Location Protocol

The rise in device mobility, interactions for service discovery as well as IPv6 networking, necessitate a means by which a clear and consistent approach towards supporting rapid, dynamic service discovery and service provision must be undertaken. This paper describes ways to provide movement detection, automatic and dynamic discovery of network services as well as network characteristics at the point of attachment for mobile devices moving in IPv6 network spaces. It employs a service discovery mechanism based on using the service location protocol (SLP), one of the most extensively researched and standardised discovery protocols in existence today. We show why traditional means of service discovery have shortcomings when used in advanced mobile networking environments, and describe new extensions to overcome these issues without losing compatibility to existing work in SLP. Prototypes developed and tested from the work described provided empirical verification over several of our production and research IPv6 networks.

Developing network software and communications protocols towards the internet of things

One of the most profound changes today is the increase in mobility of portable yet powerful wireless devices capable of communicating via several different kinds of wireless radio networks of varying link-level characteristics. This paper addresses how the design and implementation of future applications and protocols can be facilitated by network programming frameworks. For the Internet of Things, upholding a clean layered network design while explicitly taking into account host capabilities, interfaces, device resources, device associations, multihoming, user policies and context management would impose many challenges for implementation support. The paper then presents a highly interoperable and lightweight event-based framework that serves many of these needs while also modelling, monitoring and handling events intrinsically present in communication architectures.

A Real-Time Peer-to-Peer Streaming System for Mobile Networking Environment

Peer-to-peer is emerging as a potentially disruptive technology for content distribution in the mobile Internet. In addition to the already well-known peer-to-peer file sharing, real-time peer-to-peer streaming is gaining popularity. This paper presents an effective real-time peer-to-peer streaming system for the mobile environment. The basis for the system is a scalable overlay network which groups peers into clusters according to their proximity using RTT values between peers as a criteria for the cluster selection. The actual media delivery in the system is implemented using the partial RTP stream concept: the original RTP sessions related to a media delivery are split into a number of so-called partial streams according to a pre-defined set of parameters in such a way that it allows low-complexity reassembly of the original media session in real-time at the receiving end. Partial streams also help in utilizing the upload capacity with finer granularity than just per one original stream. This is beneficial in mobile environments where bandwidth can be scarce.

Simple, accurate and computationally efficient wireless channel modeling algorithm

We propose simple and computationally efficient wireless channel modeling algorithm. For this purpose we adopt the special case of the algorithm initially proposed in [1] and show that its complexity significantly decreases when the time-series is covariance stationary binary in nature. We show that for such time-series the solution of the inverse eigenvalue problem returns unique transition probability matrix of the modulating Markov chain that is capable to match statistical properties of empirical frame error processes. Our model explicitly takes into account autocorrelational and distributional properties of empirical data. We validate our model against empirical frame error traces of IEEE 802.11b wireless access technology operating in DCF mode over spread spectrum at 2Mbps and 5.5 Mbps bit rates. We also made available the C code of the model as well as pre-compiled binaries for Linux and Windows operating systems at http://www.cs.tut.fi/~moltchan.