Abstract
We present an architecture for implementing a wireless physical layer in a packet-based network simulator. We integrate this architecture in the popular ns-2 network simulator and use it to implement an impulse-radio ultra-wide band (IRUWB) physical layer. Contrary to the current wireless physical layer implementation of ns-2, in our case a packet is fully received by our physical layer before being delivered to the MAC layer. A packet detection and timing acquisition model has been implemented. Furthermore, for each packet, a packet error rate (PER) can be computed as a function of the received power, interference from concurrent transmissions, and thermal noise. This architecture is quite generic and allows for the simulation of any physical layer where an accurate model of interference is of high importance, e.g., IR-UWB or CDMA. Our implementation for IR-UWB takes into account transmissions with different time-hopping sequences (THS). The underlying modulation is binary phase shift keying (BPSK), followed by a variable-rate channel code. Our implementation is the first available that allows for the simulation of IR-UWB networks. It is modular and can thus be easily modified and extended.

Abstract
Modeling Multi-User Interference (MUI) is crucial in the design of wireless networks. In the case of Impulse Radio (IR)- Ultra Wide Band (UWB) networks, most of the adopted models are inspired by the legacy of the reference literature on spread spectrum communications, and do not address specific features for IR systems, where spectrum spreading is basically obtained by the radiation of very short time-limited pulses. The problem of conceiving a specific model for MUI in IR-UWB networks is addressed in this paper. The reference scenario consists of multiple asynchronous users transmitting IR-UWB signals using Pulse Position Modulation (PPM) in combination with Time Hopping (TH) coding. We provide a novel analytical expression for the average BER based on the observation that interference in IR is provoked by collisions occurring between pulses belonging to different transmissions. The proposed method requires specification of a similar set of system parameters as Gaussian-based approaches, but shows improved accuracy in estimating BER.