Search for: All records

Packet-level network simulators such as ns-3 require accurate physical (PHY) layer models for packet error rate (PER) for wideband transmission over fading wireless channels. To manage complexity and achieve practical runtimes, suitable link-to-system mappings can convert high fidelity PHY layer models for use by packet-level simulators. This work reports on two new contributions to the ns-3 Wi-Fi module, which presently only contains error models for Single Input Single Output (SISO), additive white Gaussian noise (AWGN) channels. To improve this, a complete implementation of a link-to-system mapping technique for IEEE 802.11 TGn fading channels is presented that involves a method for efficient generation of channel realizations within ns-3. The runtimes for the prior method suffers from scalability issues with increasing dimensionality of Multiple Input Multiple Output (MIMO) systems. We next propose a novel method to directly characterize the probability distribution of the"effective SNR" in link-to-system mapping. This approach is shown to require modest storage and not only reduces ns-3 runtime, it is also insensitive to growth of MIMO dimensionality. We describe the principles of this new method and provide details about its implementation, performance, and validation in ns-3. 

Results are presented from an extensive campaign of link simulations for multi-user multi-input multi-output (MU-MIMO) scenarios of 802.11ac wireless local area networks (WLAN) for use within a link-to-system mapping framework for ns-3 network simulation. As in [2], Exponential Effective SNR Mapping (EESM) is used inclusive of the impact of channel estimation, but this works extends beyond SISO to MU-MIMO. MATLAB® link simulation results using the WLAN Toolbox™ are used to generate an error rate table lookup for EESM to produce a corresponding packet error rate (PER) for use by ns-3. The simulation programs are made available to allow reproduction and extending of the baseline results. 

An important challenge for ns-3 is to enable efficient performance evaluation of increasingly dense and heterogeneous networks,cognizant of cross-layer (specifically, Layers 1 & 2) interactions. In this work(a continuation of U.Washington efforts),we present improved physical layer abstractions for a key component underlying all 802.11 WLAN MAC performance evaluation-the Clear Channel Assessment(CCA) procedure central to CSMA/CA-for implementation in the ns-3 simulator. We model the preamble correlation process as typically implemented in 802.11 radio and represent the resulting probability of detection as a look-up table with a parameterized correlation threshold for different receive sensitivity strategies. Further, we also added a new carrier sense threshold adjustment mechanism to allow nodes to enable bypassing the default(and to date,fixed) -82dBm threshold. Such a capability aligns ns-3 for performance evaluation of dense networks equipped with new spatial reuse mechanisms. We demonstrate this via simulation of spatial reuse gains from dynamic sensitivity control(DSC) that are verified against IEEE 802.11ax standards group contributions. Using simulation results from a fixed rate multi-BSS network,we then identify valuable design guidelines to maximize the aggregate throughput with DSC.