WLAN devices (access points and clients) supporting (up to) three spatial streams are often denoted as "3x3" devices, referring to the minimal number of transmitters and receivers required to make this particular configuration of MIMO work. But there is so much variability in terminology here that a standard nomenclature is required.
Wireless LAN devices (access points and clients) supporting (up to) three spatial streams are often denoted as "3x3" devices, referring to the minimal number of transmitters and receivers required to make this particular configuration of MIMO work. But there is so much variability in terminology here that a standard nomenclature is required.
There is, unfortunately, no universal agreement on this point. Regardless, Farpoint Group and many others have adopted the "TxR:S" convention, where T is the number of transmitting antennas, R is the number of receiving antennas, and S is the maximum number of spatial streams possible, with an 802.11n-compliant implementation of a stream limited to a PHY rate of 75Mbps in a 20-MHz. channel and 150Mbps in a 40-MHz. channel. Note here that T and R represent the number of physical antennas, not the number of radios. One radio is required on each end per stream, but antenna-diversity techniques, often used to combat radio artifacts (most notably various forms of signal fading) and applicable to both the transmit and receive sides, often result in the number of antennas being greater than the number of radios.
Cisco's 3600 AP, for example, can be denoted as 4x4:3 - still three streams, still limited to a maximum of 450 Mbps in a 40-MHz. channel, but with extra antennas on both the transmit and receive sides for purposes of diversity.
Finally, note that there need be no direct correlation between a particular antenna configuration and the Layer-7 performance of any given product, and thus such a description alone is inadequate for a full understanding of the capabilities of any given product.
Also note that architectural design and other features provide significant differentiation beyond measured throughput alone, so performance specs by themselves should never be assumed to be a key point of differentiation. Real-world testing in a facility intended for provisioning, with application-appropriate workloads, is the only case where measured performance can perform such a role.