From mboxrd@z Thu Jan 1 00:00:00 1970 Return-Path: Received: from mail.lang.hm (unknown [66.167.227.145]) (using TLSv1.2 with cipher ADH-AES256-GCM-SHA384 (256/256 bits)) (No client certificate requested) by lists.bufferbloat.net (Postfix) with ESMTPS id 511AB3B29E for ; Sat, 8 Apr 2023 13:37:38 -0400 (EDT) Received: from dlang-mobile (unknown [10.2.2.69]) by mail.lang.hm (Postfix) with ESMTP id 23557183F7C; Sat, 8 Apr 2023 10:37:37 -0700 (PDT) Date: Sat, 8 Apr 2023 10:37:37 -0700 (PDT) From: David Lang To: Ulrich Speidel cc: David Lang , starlink@lists.bufferbloat.net, George Michaelson In-Reply-To: <41ecacc1-9a34-8853-fbfe-f0e6add5e330@auckland.ac.nz> Message-ID: <145rq1sp-80n0-5q4n-sn82-364rn1n782s4@ynat.uz> References: <41ecacc1-9a34-8853-fbfe-f0e6add5e330@auckland.ac.nz> MIME-Version: 1.0 Content-Type: multipart/mixed; boundary="228850167-702620267-1680975457=:8745" Subject: Re: [Starlink] apnic piece on starlink X-BeenThere: starlink@lists.bufferbloat.net X-Mailman-Version: 2.1.20 Precedence: list List-Id: "Starlink has bufferbloat. Bad." List-Unsubscribe: , List-Archive: List-Post: List-Help: List-Subscribe: , X-List-Received-Date: Sat, 08 Apr 2023 17:37:38 -0000 This message is in MIME format. The first part should be readable text, while the remaining parts are likely unreadable without MIME-aware tools. --228850167-702620267-1680975457=:8745 Content-Type: text/plain; charset=UTF-8; format=flowed Content-Transfer-Encoding: 8BIT On Sat, 8 Apr 2023, Ulrich Speidel wrote: > On 8/04/2023 12:10 am, David Lang wrote: >> I will note that in the Starlink plans, there are plans >> to put a layer of satellites at a sigificantly lower altitude. > I should add to this that this would seem like a good strategy, except of > course that this comes with its own set of challenges. Earth observation > satellites in particular are abundant in lower orbits - if you have a camera > on board, you want to be as close to your subject as you can. So there isn't > quite as much space down there as there is further up. > > Residual atmospheric drag at lower altitudes is also higher, which means you > either need to take more fuel to compensate (=heavier satellite & fewer sats > per launch) or you need to replace the satellites more often. > > You also need more satellites for global coverage in a shell like this. > > Also, as you mostly look at satellites sideways when you're a ground station, > the path length and therefore the path loss isn't necessarily all that much > lower - going from 550 km to 275 km gives you an extra 6 dB of gain if the > satellite is straight overhead, but that advantage shrinks as you move away > from zenith. the published plan is a shell at ~340km (7500 satellites) in addition to the one at ~550km (and a possible 'long haul' shell at ~750km with <1k satellites) >> By launching 10x as many satellites, and each one being able to handle 10x >> the data, they _may_ get to 100x, but that is really going to be pushing it. >> (note that this is for ~10x the number of satellites lauched by everyone >> other than SpaceX since Sputnik) > > Having seen figures of ~45k sats bandied around for some proposed > mega-constellations, the 10x number of satellites might just work out. regulatory approval required, but starlink is aiming for ~42k satellites, they are only approved for somewhere around 10-12k so far and I believe they are nearing 4k in orbit. > Whether we'll get to 10x the capacity per satellite is another question > altogether given spectral constraints. yes, to get to this sort of capacity, you need to have multiple satellites covering each cell at a given time, which per the research published not that long ago was not yet the case. > With ISLs, one could in principle free > up part of the gateway traffic spectrum by putting gateways in areas that are > devoid of other users, but quite how practical that is given that remote > areas are where LEOs will be needed most is a good question. supporting terminal-to-terminal traffic also opens interesting possibilities (although less than we would like due to the server-centric nature of current Internet usage) > One option that could push things a little further in conjunction with LEOs > would be HAPSĀ  - high altitude platform systems, essentially solar-powered > UAVs that act as stratospheric cell towers with tours of duty measured in > weeks or months. These could use lasers as backhaul to LEO networks, yet > project comparatively narrow phased array beams to users on the ground. A > HAPS flying at 30 km overhead has a path loss that's around 25 dB below that > of a LEO sat at 550 km, and a clear optical path to the satellites above. > Technology isn't quite there yet - essentially, we're at the point where > solar cells have become performant enough in conjunction with batteries that > have become light enough to allow sustainable cyclic recharging of a UAV's > flight systems. But there are still issues to be addressed around the excess > power required to operate a cell site in the sky and or course all the > regulatory and safety aspects associated with operating things that don't > burn up when they come down. Google shuttered project Loon (balloons to do this rather than UAVs) in the last year or two >> If you can get fiber, it's always going to be better than a wireless option, >> DSL is threatened by Starlink in many suburbs, cablemodems depend so much on >> the ISP it's hard to say > > This is an interesting comment. Completely agree on the fibre aspect. > > DSL I think is threatened more by fibre than Starlink in most places (except > the US perhaps), which has basically displaced most DSL connections where it > became available. We were on DSL here till 2017, and as fibre was on the > horizon for a while, the company that runs the cable network here on behalf > of the telcos stopped investing in new DSLAM modules, instead preferring to > switch customers with problematic ones to modules that had become available > as a result of customers migrating to fibre. We found ourselves with a weird > problem literally overnight one day - intermittent disconnects lasting a > minute or two. These persisted through a change of DSL router, and logging > these for a few days showed a clear diurnal peak time pattern - so it was > obvious we were dealing with DSLAM-side crosstalk issues here. I asked to be > switched to a different DSLAM. This was an odyssee of support calls given > that you cannot call the lines folk directly - you must call your retail ISP, > who pays someone in India a few rupees to tell you to reboot your router to > make the problem go away. By the time I'd educated their 3rd tier support > about what crosstalk was, I'd literally spent many many hours on the phone to > India. Eventually, they switched me over to a new DSLAM and the problem went > away for a few months, just to return as they kept rewiring more legacy > customers. when I switched from cablemodem to 129k SDSL in 2002, my usable bandwidth significantly improved. > For those who still have DSL now, VDSL plans start at less than half of what > Starlink charges, with potentially comparable data rates, so not everyone > will want to switch. I live in the greater LA area, in the middle of a town of >125k The best DSL I can get takes two phone lines to give me 8m down 1m up (in theory I should get 10/2 but the line quality doesn't support it), and this costs me significantly more than starlink does. > Where I see uptake of Starlink in urban areas here is by (a) geeks and (b) > folk who want a (secondary) connection that is independent of local telcos > that run inane call centres in India. As more people work from home, just having a secondary connection becomes more important. > Cable TV and cable modems are of course pretty much unheard of here - are > there any cable modems / ISPs that do more than a few dozen Mb/s down? yes, cable modems can push 1G. I have one at 600/30 (but as a business line with static IP addresses, it costs me about triple the starlink connection) David Lang --228850167-702620267-1680975457=:8745--