As the use of LTE over unlicensed spectrum moves toward field trials expected later this year, LTE-U testing results that show mixed results as to its ability to co-exist fairly with Wi-Fi have been presented by a number of industry players. Even the basic concept of what constitutes “fair” co-existence is in the eye of the beholder.
The studies that have been done are hardly apples-to-apples comparisons, and both sides are calling the others’ test methodologies skewed toward their preferred results, particularly in recent comments to the Federal Communications Commission. Adding to the challenge is the fact that LTE-U specifications developed by the industry consortium LTE-U Forum involve proprietary mechanisms (such as Qualcomm’s Carrier Sensing Adaptive Transmission), and the version being standardized by 3GPP, Licensed Assisted Access, is substantially different in that it incorporates a mandatory Listen Before Talk mechanism required in areas such as Japan and Europe.
Still, let’s take a look at the some of the testing that has been done thus far, and what different companies have concluded based on their evaluations. Today we’ll see what Qualcomm, Broadcom and Google have said, and leave the LTE-U Forum and CableLabs for a second installment coming tomorrow.
Qualcomm: Perhaps the most frequently cited by proponents of LTE-U, Qualcomm has done testing in support of the work of the LTE-U Forum, as well as some recent additional testing to address comments by other companies to the FCC.
Qualcomm has done a series of tests and demonstrations using eight Wi-Fi routers and gradually changing nodes which were Wi-Fi and which were LTE-U, and found that, in general, the performance of Wi-Fi in the presence of LTE-U was slightly better than in the presence of another Wi-Fi node. Details of that testing have been presented by the company in a number of forums – the presentation at the recent ISART conference by Yongbin Wei, senior director of engineering for Qualcomm Technologies, summarizes its testing for both LTE-U and LAA (available for download on this page).
In its most recent round of test results submitted to the FCC, Qualcomm argued that it is unfair to compare Wi-Fi’s performance in an interference-free environment to its performance in the presence of LTE-U, as has been done in some testing. Qualcomm said that a more fair comparison is to Wi-Fi’s performance in the presence of other Wi-Fi nodes – in both instances, throughput tends to drop precipitously compared to a single Wi-Fi device operating in a clear channel.
It conducted its tests in a radio frequency isolation chamber, over the air, and said the tests were “specifically designed to replicate (to the fullest extent possible) the test scenarios cited in Google’s FCC filing.” More on that later. Qualcomm tested Wi-Fi devices from four vendors and in general, the numbers for Wi-Fi/LTE-U co-existence were comparable to Wi-Fi/Wi-Fi co-existence, particularly in downlink testing.
One thing Qualcomm’s testing did make clear, however, is that different vendors will be impacted quite differently in the presence of LTE over unlicensed. For instance, in uplink testing done on devices from four different vendors, in two cases there was less than 1 Mbps of speed difference when a Wi-Fi node was operating with another Wi-Fi node vs. an LTE-U node; in another case, the average uplink throughput was nearly 9 Mbps slower in the presence of LTE-U than other Wi-Fi, and in a fourth case, the Wi-Fi node was 30 Mbps faster in the presence of LTE-U than a second Wi-Fi node (see chart on appendix page 12 in this document).
The company has a general synopsis of its testing and demo videos available here.
Broadcom: Broadcom has been active in both comments to the FCC and in work in standards bodies; it is not part of the LTE-U Forum.
In its FCC comments, Broadcom cited simulations it has done as showing that, with proper and robust co-existence mechanisms, it’s possible for LTE over unlicensed spectrum to be a good neighbor to Wi-Fi – but much depends on implementation. Broadcom reported that in simulations with four Wi-Fi APs in one network, and four LAA nodes in a second network (using EU regulations), with 10 data users on both networks, that both throughput and latency were negatively affected by LAA if only the baseline regulations were followed. With more robust co-existence mechanisms, Broadcom found that average Wi-Fi speeds did increase slightly in the presence of LAA (34 Mbps vs. 33 Mbps with both networks on Wi-Fi) and that latency was essentially the same (44 milliseconds with two Wi-Fi networks, 43 ms with one network operating on LAA). In order to achieve those results, Broadcom recommended that LTE over unlicensed implementations have the ability to detect weak Wi-Fi signals (-82 dBm), an initial wait time of at least 43 microseconds, and exponential back-off.
Google: In its comments to the FCC, Google included a co-existence study by Nihar Jindal and Donald Breslin on LTE-U and Wi-Fi. In testing of LTE-U, the authors noted that since LTE-U equipment was not available, they generated LTE-U transmissions with a signal generator, “using the description of LTE-U coexistence mechanisms contained in LTE-U Forum documentation.” Tests were conducted in an RF isolation chamber with programmable attenuators, with single Wi-Fi AP-client pairs running TCP or UDP sessions and a single LTE-U eNodeB. Only LTE-U transmissions in the unlicensed bands were considered, and LTE-I client devices were not emulated.
Google’s study found two primary issues with LTE-U: Duty-cycling, or the on-off periods in which LTE-U and Wi-Fi take turns operating, effectively means that LTE-U steps on Wi-Fi’s toes by starting transmission before Wi-Fi is finished, leading to corruption of frames that can reduce throughput and increase latency; and that “LTE-U does not have an effective coexistence technique to handle scenarios in which LTE-U and Wi-Fi devices hear each other at moderate (below -62 dBm) power levels and, as a consequence, Wi-Fi can be crippled in such scenarios.”
In terms of LAA, Jindal and Breslin concluded that “the use of LBT could create the pathway for LAA to coexist gracefully with Wi-Fi and other unlicensed technologies, but only with very careful study of coexistence issues (coexistence studies are a part of LAA) in close collaboration with the broader unlicensed community, e.g. IEEE 802.11 and the Wi-Fi Alliance.”
Stay tuned for part two.
