Summary Distributed denial of service (DDoS) attacks have for the last two decades been among the greatest threats facing the internet infrastructure. Mitigating DDoS attacks is a particularly challenging task as an attacker tries to conceal a huge amount of traffic inside a legitimate traffic flow. This article proposes to use data mining approaches to find unique hidden data structures which are able to characterize the normal traffic flow. This will serve as a mean for filtering illegitimate traffic under DDoS attacks. In this endeavor, we devise three algorithms built on previously uncharted areas within mitigation techniques where clustering techniques are used to create geographical clusters in regions which are likely to contain legitimate traffic. We argue through extensive experimental results that establishing clusters around this narrative is a superior solution to clustering algorithms which rely on bitwise distances between IP addresses. In addition, the DDoS filtering algorithm is deployed in a virtual Linux environment using Nfqueue and tested in a simulated real‐life DDoS attack.

This paper considers a wireless multiuser system with N users that can transmit over K orthogonal frequency bands. In this system, interference from other users is treated as noise. In this system we assume each user has a target rate and the goal is to allow as many users as possible to achieve their target rate. We propose a new power allocation algorithm that minimizes both number of frequency bands to be used and necessary power at any user to achieve a target rate. We show that this approach reduces the overall interference power in the system, and thus increases the number of users achieving a desired target rate compared to other power allocation algorithms. This is supported by simulation results, which also shows that in a dense network, sum capacity of the system is increased compared to regular water filling.

This paper considers coordination of users in wireless ad-hoc networks. We define coordination as a scheme that tries to achieve the rate region given by Shannon's capacity by allowing users to exchange information and agree upon future actions, while still treating interference as noise. This differs from cooperative schemes which try to increase the achievable rate region by introducing and exploiting structure in the interference. The advantage of coordination compared to cooperation is that coordination does not require strict user synchronization and global knowledge of system parameters. Using orthogonalization as the means of coordination, we find conditions for when users should coordinate. Through simulations we investigate the performance of coordination compared to selfish and selfless autonomous power algorithms. Results show that in some scenarios coordination yield high gain, however since our coordination scheme only guarantees to outperform selfish algorithms; full orthogonalization can outperform our coordination scheme in some scenarios.

The performance of a turbo code can be severely degraded if no trellis termination is employed. This paper investigates the implications of the choice of trellis termination method for turbo codes, and explains the origin of the performance degradation often experienced without trellis termination. An efficient method to derive the distance spectrum of turbo codes for different trellis termination methods is presented. Further, we present interleaver design rules that are tailored to each termination method. Using interleavers designed with these restrictions, we demonstrate that the performance difference between various termination methods is very small, including no trellis termination at all. For example, we demonstrate a turbo code with a 500-bit interleaver that exhibits no sign of an error floor for frame error rates as low as 10/sup -8/, even though no trellis termination is employed.

The performance of a turbo code is dependent on two properties of the code: its distance spectrum and its suitability to be iteratively decoded. Both these properties are dependent on the specific interleaver used in the turbo-coding scheme. This paper describes a strategy for interleaver design that includes a criterion for the performance of iterative decoding. This criterion is based upon correlation properties of the extrinsic information, which is used as a priori inputs to the constituent decoders. Simulations comparing interleaver choices indicate an improved turbo code performance for interleavers designed with algorithms including a correlation criterion.

It is essential to find suitable modes (combinations of alphabet sizes and code rates) for transmission in cellular systems. Bad choices result in low spectrum efficiency. An analytical method is proposed to find the modes which give the highest spectrum efficiency when using adaptive modulation. It is demonstrated that data rates considerably higher than 271 and 384 kbps are feasible in GSM and EDGE systems by using adaptive modulation. It is further demonstrated that the spectrum efficiency can be improved more than 100% by using three modes in a GSM system.

A new soft decision criterion is derived. This is useful in communication systems where the combination of modulation, interleaving, and error correction is used. The new algorithm is attractive to implement since traceback techniques can be used and only a little extra memory is required to obtain the soft information associated with every symbol. Furthermore, the operations involved are not recursive and the execution is time independent of the received signal quality. The performance of the new algorithm is typically within 0.25-0.1 dB (bit error rate (BER)=10/sup -1/) of that of a significantly more complex algorithm reported previously.

A statistical multipath propagation model based on the energy distribution within discrete-time delay bins is presented. The physical reason for modelling the multipath channel by energy ratios instead of distinct ray components is the limited time resolution inherent in all channel sounders used to measure the impulse response. The difficulties in resolving the real multipath components are verified by simulations and comparisons are made with measurements. The normalized energy ratios within discrete-time bins are calculated from measurements made in different indoor environments and comparisons show that the propagation conditions have a strong influence on the energy distribution.

The bit-error-rate of a CDMA system using two types of RAKE receivers are calculated considering random codes, AWGN and a two ray multipath channel. The modulation scheme is binary PSK. The processing gain is a parameter. For low bitrates it is not very important which of the two RAKE receivers is used, but when the bit-duration approaches that of the channel, it is very important that the RAKE receiver is properly implemented.

The autocorrelation function of a binary multi-hcoded signal is determined analytically. This is done by considering the probability density of the "phase change variable" and then deriving the characteristic function. From the autocorrelation, the power spectrum can be found by an analytical or numerical Fourier transformation. Two examples are given.

Some properties of digitally phase modulated (DPM) signals are presented. Phase modulation with (overlapping) pulses generated by a digital FIR filter belong to this class, which may be considered to be a practical approximation to continuous phase modulated (CPM) signals. The power spectra of DPM signals are derived analytically. The ability of these signals to operate through an additive white Gaussian noise channel is assessed by calculating their minimum Euclidean distance. Their noise and spectral properties are found to be similar to those of CPM signals. Assuming that a Viterbi decoder is used to resolve symbol interference, the out-of-band power tends to decrease as the pulse duration increases, and the noise immunity is enhanced. At the same time the receiver complexity grows exponentially. Hence, noise immunity and spectrum compactness are achieved at the cost of higher received complexity. Modems for DPM signals are believed to be easier to implement than those for CPM signals. This is because filter design is simple and a residual carrier component can be retained to facilitate carrier regeneration. Furthermore, the accumulated carrier phase does not need to be continuously evaluated in order to perform matched filtering. The analytic results derived are supported by measurements and simulations.

The output EIRP or the available bandwidth of a satellite in general limits the amount of traffic which can be handled by SATCOM systems. As the frequency congestion drives future SATCOM systems to higher frequencies, the former constraint will be the more important one because of the severe fades caused by precipitation. Rather than using fixed fade margins, the link power can be dynamically allocated from a common pool of satellite EIRP. By judicious, adaptive allocation of the satellite EIRP among the various users, the traffic supported can be maximized while maintaining the link availability. This paper reports techniques and statistical results of four different ways of allocating the EIRP. By using adaptive centralized control of the terminals transmit power and a variable satellite transmit antenna adaptively controlled by ground commands to compensate for link fades, most efficient use of the satellite EIRP is made. The performance of this technique is derived and compared with others of less complexity and that of a system with fixed satellite antenna and transmit power levels.

A dynamic mathematical model of rain attenuation has been developed and is presented in this paper. This model permits the expression of analytic relationships between parameters commonly used to describe the properties of interest for communication. The dynamic model is based on the lognormal distribution of rain attenuation and utilizes a memoryless nonlinear device to transform attenuation and rain intensity into a one-dimensional Gaussian stationary Markov process. Hence, only one parameter is required to introduce the dynamic properties of rain attenuation into the model. Experimental results and the known properties of rain have been used to derive and to verify the model; comparative results are presented and demonstrate good correspondence. The application of the model to the statistical analysis of the performance of communications systems is illustrated in the paper. The use of a dynamic rain attenuation model is necessary in order to analyze radio communication systems with transmit power control to offset the effects of rain attenuation, and where the finite response time of the control system affects the performance. An advantage of the model is the simplicity with which it allows simulation of communication link performance under the influence of rain attenuation. Such simulations are of great interest for complex models of adaptive networks where several deteriorating effects, including finite response times, are present.

The code division multiplexing (CDM) of digitally modulated baseband signals is normally referred to as majority logic multiplexing (MLM). The employment of MLM techniques in spread spectrum multiple access (SSMA) systems is especially attractive because it allows many different types of baseband channels to be easily combined. The traffic level can be gradually reduced to obtain a desired increase in processing gain, without affecting the active channels' share of the transmitter power. The code integrity and the processing gain of the SSMA equipment can be preserved by suitable design. This paper contains a general performance analysis of an MLM system that uses random codes. The basic MLM principle is extended to handle the combining of traffic channels with different bit rates. Calculated values of performance deterioration due to noise are compared with experimental results and found to be m close agreement.

The characteristics of a bandpass nonlinearity are expressed in this paper by a finite series expansion of the complex envelope transfer function. This function is commonly acquired by direct measurement of the single-carrier envelope characteristics. This paper describes an alternative method whereby the complex envelope transfer function is obtained by measuring the power of the intermodulation products generated by two-tone signal inputs. This approach is known to permit a more accurate characterization than the other method, particularly in the cases of mild nonlinearities. The fundamental problem that has to be overcome in order to establish the complex envelope characteristic from power measurements alone, is the determination of the phase of the measured intermodulation components. The solution, as developed in this paper, is to measure the two-tone spectrum at two power levels when the nonlinearity only exhibits amplitude distortion, and at three power levels when the nonlinearity exhibits AM-to-PM conversion in addition to amplitude distortion.

A degenerate parametric amplifier can be shown to have phase-regenerative effects on a phase-shift keyed (PSK) signal corrupted by noise. In this paper the stationary output-phase density is calculated when the input signal consists of a two-phase PSK signal and white Gaussian noise.