CTU: Capturing Throughput Dependencies in UWB Networks Ioannis Broustis ∗ , Angelos Vlavianos ∗ , Prashant Krishnamurthy  , Srikanth V. Krishnamurthy ∗ ∗ University of California, Riverside,  University of Pittsburgh {broustis, aggelos, krish}@cs.ucr.edu, prashant@mail.sis.pitt.edu Abstract—The inherent channel characteristics of impulse-based UWB networks affect the MAC layer performance significantly. Previous studies on evaluating MAC protocols are based on prolonged simulations, and do not account for the multiple-access interference that arises due to multipath delay spread. In this work, we develop CTU, an analytical framework that captures the performance of MAC protocols, while taking into account the underlying PHY layer effects. The key attributes that make CTU novel are: (a) It is modular and therefore flexible; it can be easily modified to provide a basis for characterizing and evaluating a wide range of MAC protocols designed for impulse-based UWB networks. The only requirements are that the MAC protocol under study be based on time-hopping, and the modulation scheme be pulse position modulation; these are common design decisions in most impulse based UWB networks. (b) It considers the channel characteristics in addition to MAC layer effects; in particular, CTU correlates probabilistically the multipath delay profile of the channel with the packet error rate. We employ CTU to evaluate the performance of a generic medium access procedure. We compare the results with those from extensive simulations and show the high accuracy of CTU. We use CTU to assess the impact of various system parameters on the MAC layer performance; we make several interesting observations that are discussed in depth. Index Terms—Ultra-Wide Band (UWB) 1 , Wireless Communica- tions, Multipath Delay Spread, Modulation. I. I NTRODUCTION The medium access performance of multi-hop impulse-based UWB networksis significantly affected by the PHY layer char- acteristics, and in particular by the multipath delay spread [1]. Due to reflections from various objects, multiple copies (rays) of the same transmitted signal appear at the receiver; each copy has a different amplitude, phase and delay. The copies interfere with later pulse transmissions and this significantly affects the long-term network throughput. When evaluating MAC protocols designed for UWB networks, it is critical that one accounts for these effects. Previous studies on UWB, however, are limited to either (i) possibly time-intensive simulation-based MAC protocol evaluations (most of such efforts do not account for multipath delay spread effects), or (ii) PHY layer analyses to model the behavior of a single link in the presence of delay spread. A cross-layer analytical framework that quickly and accurately quantifies the impact of the UWB PHY layer attributes on MAC layer mechanisms, could be of great help throughout the protocol design process. In this paper, we develop an analytical framework to compute the average saturation throughput and evaluate its dependencies on PHY layer effects, in multi-hop UWB networks. The sat- uration throughput is defined to be the throughput when the nodes in the network always have packets to send. We call our framework “CTU” for short (for Capturing Throughput dependencies in UWB networks). The key attributes of CTU are: (a) it is modular and therefore flexible, and (b) it captures 1 This work is supported in part by the NSF CAREER Grant No. 0237920 and the NSF NRT grant No. 0335302. both physical layer characteristics and MAC layer effects. CTU computes the probability of packet loss due to (i) MAC layer collisions and (ii) multiple access interference (MAI) due to delay spread effects. It then combines them to compute the sat- uration throughput. The immediate benefit from our analysis is that it obviates the need for repeated simulations. We elaborate on the attributes of CTU below: CTU is modular and hence applicable to a wide set of MAC protocols. The part of the CTU framework that computes packet losses due to MAI is generic and can be used in con- junction with any MAC layer protocol. The only requirements are that the protocol be based on time-hopping (TH), a multiple access procedure, and that the underlying modulation scheme be Binary Pulse Position Modulation (BPPM); this is common to most PHY/MAC layers designed for UWB networks. The building block for CTU is a basic MAC layer procedure that is likely to be the foundation for more complex TH-based MAC layer protocols. CTU accounts for cross layer dependencies. Multiple- access interference in UWB networks not only depends on the extent of delay spread but also on the access procedure defined by the MAC protocol. CTU provides a seamless methodology of integrating the access procedure with the effects of delay spread; in other words, it captures the cross layer dependencies between the PHY and the MAC layers. In particular, CTU incorporates the impact of delay spread on the multiple access performance based on a modified version of the Saleh-Valenzuela (SV) model that has been adopted by the IEEE 802.15.3a task group [2], (discussed in more detail in section II). Evaluation of CTU. We apply CTU on a simple MAC protocol, and we compute the per node throughput. We validate the accuracy of our analytical results, by comparing them with results from extensive simulation experiments that we perform. We also find that while obtaining simulation results takes days, generating results with CTU requires less than one hour. Using CTU to assess system performance. With CTU, we examine the impact of various system parameters on the satu- ration throughput. We discuss in detail, many interesting trends that are observed. We believe that CTU can help designers and implementers choose UWB MAC layer system parameters appropriately during network design. The remainder of the paper is structured as follows. In section II we provide background on the UWB PHY layer and discuss related work. In section III, we describe our analytical frame- work in detail. In section IV, we compare the simulated behavior of the candidate MAC protocol with the behavior observed with CTU. We also describe the impact of various system parameters on network throughput under saturation conditions, report the trends that we observe and the interpretations thereof. In section V we discuss the scope of CTU in terms of its tunability and its limitations. We conclude in section VI.