Our client, a Fortune 100 technology company, wanted to simulate wireless performance and compare two different modes of operation for the purposes of research and development. They owned a single piece of spectrum, and wanted to determine which of two competing topologies would have a better performance.
Manifold built a multi user simulation of the 802.11AF MAC (medium access control) layer to answer their question. We synthesized the raw results from simulation to make a recommendation to the client about which approach they should pursue.
Manifold worked with the client to thoroughly understand their business needs and then determined which tools were readily available to help solve their problem and which would need to be built. We tapped our extensive expert network regarding the wifi MAC layer, and worked with an expert to develop a state machine diagram appropriate for the wireless protocol.
Wifi is an uncoordinated wireless network, which means that individual devices don’t coordinate with each other when sending packets of information across the air. If a device (or “node”) attempts to talk to the access point at the same time that another node it can’t see is already talking, then it will cause packet collisions—this is called the “hidden node problem.”
We used a Monte Carlo simulation—a multi-agent simulation where the devices (or “agents”) are placed and interact with each other through this state machine. We needed to simulate multiple internet-connected wireless devices on the network, each of which may or may not be able to see each other. The Monte Carlo simulation allowed us to place these devices randomly, and their position determined to what degree they could see each other. We ran these random simulations 200 times, and used the resulting data to determine a set of average behaviors.
We used an off-the-shelf PHY (physical layer) radio propagation model called the Hata model to determine how the signal attenuation drops off as the distance between devices changes. Using the Hata model inside the Monte Carlo simulation allowed us to determine with a high level of precision to what degree a given pair of devices would be able to see each other. When they weren’t able to see each other, then we knew to expect a high degree of packet collision.
Manifold implemented the simulation in MATLAB, and ran the simulation for various parameters. We then presented the results to the user as an Excel tool that allows the user to change the parameters and receive answers computed from raw simulation results.
Manifold made detailed evaluations of the two approaches being considered by our client, and made a definitive recommendation as to the best option. As it turns out, we showed that the performance they were hoping to obtain was ultimately not possible in either simulation. This information provided a valuable level-check on their expectations and helped them to revise their business plans accordingly.