Wednesday, January 02, 2019

Simulation of OneWeb, SpaceX and Telesat's proposed global broadband constellations

Inigo del Portillo and his colleagues at MIT have run a simulation comparing OneWeb, SpaceX and Telesat's proposed low-Earth orbit Internet service constellations. The models are based on the initial FCC filings by the companies and demand and data-rate estimates by the authors. (I will mention subsequent amendments to the filings below).

The first part of the paper presents an overview of the satellites and constellations and simulations predicting the average number of satellites within line of sight at different latitudes:

Source

With 4,425 simulated satellites, SpaceX is the clear leader in all but the extreme north and south, but Telesat, with only 117 satellites, averages more visible satellites than OneWeb with 720 because their minimum elevation angle is 20 degrees while OneWeb's is 55 degrees (SpaceX's is 40 degrees).

Note that SpaceX coverage peaks around the latitude of Seattle, where their center for satellite operations is located, and many large trading and financial cities have 30 or more visible satellites:

Source

They also simulated alternative ground segment configurations. The following figures show throughput as a function of the number of ground stations and assumed optical inter-satellite link (OISL) speed:

Source

They estimated the average data rate per satellite and total system throughput (sellable capacity) for each constellation then computed the number of ground stations needed to achieve full capacity:

SpaceX OneWeb Telesat
Average data rate per satellite (Gbps) 5.36 2.17 22.74
Total system throughput (sellable capacity, Tbps) 23.70 1.56 2.66
Ground stations needed to achieve full capacity 123 71 42
Throughput with 50 ground stations (Tbps) 16.80 1.47 2.66

Note that marginal improvement diminishes as the number of ground stations increases. For example, with 50 ground stations each, the throughputs are SpaceX 16.8, OneWeb 1.47 and Telesat 2.66 Tbps.

Since SpaceX has their own launch capability, they can afford to begin at very large scale, giving them a significant capacity advantage, but Telesat has a significant lead in capacity per satellite. Putting these constellation capacities in context, Telegeography reported that global International bandwidth in 2018 was 393 Tbps, two-thirds of which has been deployed since 2014. Satellite constellations with OISLs will add to that total and provide lower latency times than international terrestrial cables.

As mentioned above, there have been some changes in plans since the FCC filings upon which these simulations were based. OneWeb has reduced the number of satellites in their constellation to 600 and eliminated ISOLs and SpaceX has lowered the altitude of their constellation from 1,100 to 550 km and cut the number of ISOLs per satellite from 5 to 4. (SpaceX was also authorized to use V-band spectrum, but that may be for their planned VLEO constellation).

I asked del Portillo whether they had rerun the simulations using the revised plans, and he said they had not, but he did not think that the numbers for total throughput and number of ground stations would vary a lot for SpaceX and he expected the total system throughput would decrease slightly for OneWeb because of the reduced number of satellites. He also expressed concern about OneWeb being able to provide continuous coverage near the equator where there might be coverage gaps or lower data-rates. (Note that the simulation focused on end-user and backhaul demand, not marine and in-flight connectivity which are priority markets for OneWeb).

SpaceX appears to have financing in place to cover their initial launches and achieve high capacity early and Telesat is making a much smaller initial investment, but should be able to scale up as demand increases. Based on this simulation and problems they have encountered with their Russian partners, OneWeb may be falling behind.

I've summarized some of the results of these simulations, but you should read the paper for details on the model structure, assumptions and a discussion of technical challenges. You can also check out this slide presentation on the paper. I also recommend an earlier paper and accompanying videos on a simulation that predicts routing redundancy and latency with SpaceX's planned broadband constellation.

Wednesday, December 26, 2018

Low-Earth orbit (LEO) satellite internet service developments for 2018

At the end of 2017, I posted a review of important LEO-satellite Internet service developments. I've been updating those posts during the year and have also added the following new posts this year:

OneWeb's offer to sell a share of their satellite Internet service to Russia. (December 2018).

OneWeb has a Russian contract to launch Internet-service satellites and a marketing partnership with a Russian company. When the Russian government expressed concern about security and US sanctions, OneWeb made concessions. It was reported that OneWeb had offered to sell 12.5% of the company to Russia, giving them a board seat and access to technical information, but OneWeb denied the report. Given Russia's accelerated investment in their global satellite navigation system, history of hacking the US Global Positioning System and using the Internet to foment social unrest and influence election results in Europe and the United States, I would be reluctant to share technical information with them.

A cool simulation of SpaceX's revised satellite broadband plan. (November 2018)

On November 15, the FCC approved a revision to the plan for Starlink, SpaceX's forthcoming broadband satellite service. The new plan reduces the number of satellites from 4,525 to 4,409 and lowers the altitude of the phase-1 satellites from 1,100 to 550 km. Mark Handley, a professor at University College London, has created videos based on simulations he ran of both the original and revised phase 1 plans. The videos and his narration provide insight into the assumptions he made in building his models and constellation design decisions and tradeoffs.

An in-depth interview of OneWeb Executive Chairman Greg Wyler. (November 2018)

In this "definitive" 2018 interview of OneWeb founder and Executive Chairman Greg Wyler, he said their initial customers will include emergency services, mobility (aviation), health centers and schools. He did not mention individual homes or the problems he has been having negotiating with the Russians or dropping inter-satellite laser links. There is a lot more -- check the interview out.

Obstacles in OneWeb's negotiations with Russia. (October 2018)

OneWeb has a contract to launch satellites on Russian-made Soyuz rockets and has formed a joint venture with Gonets, a Russian satellite operator, as a marketing partner. Recently, Russia has expressed concerns about security, drug their feet on spectrum allocation and complained about US sanctions. Subsequently, OneWeb reduced its interest in the Gonets partnership from 60 to 49 percent, giving Gonets control. This case illustrates the fact that political, security and financial negotiations may be as difficult as designing satellites and rockets for a would-be global Internet service provider.

Might CubeSats provide broadband Internet connectivity one day? (September 2018)

Cubesats are small, standardized satellites. Since they are small, they are relatively cheap to launch and standards and standard components make them relatively cheap. This post looks at the efforts of two startups working on narrowband communication applications using CubeSats. Today's CubeSats are smaller and less powerful than those planned for Internet constellations from companies like OneWeb, Telesat or SpaceX and it's hard to believe that the capability of broadband Internet-service satellites might one day fit in a CubeSat -- but think about the phone in your pocket.

SpaceX Starlink test -- good news but unanswered questions. (June 2018)

Elon Musk tweeted that the two SpaceX test satellites, TinTin A and B, are connecting at "high bandwidth" with 25 ms latency. That's good news but it leaves a lot unanswered. For example, he did not mention the speed and reliability of the phased-array handoffs between the satellites and ground terminals and he said nothing about tests of the inter-satellite laser links, which OneWeb has abandoned.

SpaceX President and COO Gwynne Shotwell on synergy among Musk companies and Starlink profit. (May 2018)

In a recent interview, SpaceX President and COO Gwynne Shotwell said that SpaceX is profitable, but she predicts that Internet service revenue will soon be much greater than that of their launch business and the Wall Street Journal agrees. She also pointed out synergies between Elon Musk's companies and said they remain on schedule to take people to Mars in 2024.


Telesat begins testing low-Earth orbit satellite Internet service
. (May 2018)

Telesat's demonstration satellite is now ready for testing with maritime connectivity provider OmniAccess, Australian ISP Optus and in-flight entertainment company Global Eagle Entertainment. They are moving quickly and their initial focus is on these specialized markets.

Elon Musk tells what to expect from the Block 5 Falcon 9 rocket. (May 2018)

On the occasion of it's first production flight, Elon Musk said this will be the last major version of the Block 9 Falcon 9 before their next rocket, the BFR. The Block 5 is designed for rapid-turnaround reusability and he expects there to be 300 or more Block 5 flights before it is retired. The rockets are designed to do 10 or more flights without refurbishment and expected to be capable of at least 100 flights before being retired.

FCC approves SpaceX Starlink's Internet-service constellation -- now there are four. (March 2018)

SpaceX, OneWeb, Telesat and Space Norway (focused on the far north) now have FCC approval to provide fixed-satellite service using constellations of LEO satellites.

O3b satellite Internet -- today and tomorrow. (March 2018)

SpaceX, OneWeb and Telesat are planning to offer Internet-service from LEO, but O3b is already providing connectivity to relatively large customers like mobile phone companies, government organizations, and cruise ship lines using a constellation of medium-Earth orbit (MEO) satellites. For example, they serve mobile phone company Digicell in Papua New Guinea and Cuban ISP ETECSA. O3b plans to add four more satellites early next year and will begin deploying their next-generation constellation, mPower, in 2021. While today's O3b satellites have 10 steerable edge-terminal beams, the mPOWER satellites will have over 4,000 steerable beams that can be switched under program control giving the initial seven-satellite constellation over 30,000 dynamically reconfigurable beams and over 10 Tbps capacity.

The current and future role of satellite Internet service in Cuba. (12 posts)

Today Cuba makes limited use of geostationary and MEO satellite connectivity, but LEO and MEO satellite constellations could play a major role in future Internet connectivity there and in other developing nations.

OneWeb's offer to sell a share of their satellite Internet service to Russia

Update: OneWeb has denied that they have offered to sell any stake in the company to the Russian Government and Reuters has now reported that Russia’s communications minister said that "the government has not discussed such a deal with OneWeb."

OneWeb confirmed that they are restructuring their existing commercial joint venture with Russian partner Gonets, to comply with certain Russian regulatory requirements. They reiterated that "the joint venture is solely for the commercialization of OneWeb’s satellite broadband services to customers in Russia" and stated that the joint venture would "not have access to OneWeb's satellite technology or related know-how."
-----

Given the context, I am not keen on OneWeb sharing technology with Russia before or after a deal is closed.

Number of global satellites (source)
OneWeb has a contract to launch Internet-service satellites on Russian-made Soyuz rockets and has a joint venture with Gonets, a Russian satellite operator, as a marketing partner in Russia. However, the Russian government has expressed concerns about security, drug their feet on spectrum allocation and complained about US sanctions. This led OneWeb to make two concessions -- they dropped inter-satellite laser links from their constellation design and they cut their interest in the Gonets joint venture from 60 to 49 percent.

Apparently, that did not satisfy the Russians. Reuters reports that according to three sources, OneWeb has offered the Russian government the option of buying a 12.5-percent stake in the company in exchange for approving its request for a frequency band.

If Russia buys a minority stake in OneWeb, it would be represented on the company’s board of directors and have access to the project’s technical documentation. One of the sources -- a Russian government official -- said the government was not satisfied with the conditions attached to the purchase -- it wanted access to the technical documentation before committing to the deal.

This offer needs to be put in context:
    Active disruption by Russia (source)
  • There are four competing global satellite navigation systems: GPS (US), Galileo (Europe), GLONASS (Russia) and Beidou (China).
  • University of New Brunswick professor Richard Langley, who has worked extensively with the Global Positioning System since the early 1980s says that "while the GLONASS service level has dropped below acceptable levels, it is now a dependable system and, with announced improvements, will be a contender in the future world of multi-GNS.
  • Russia has been accused of jamming and spoofing the GPS.
  • Russia has used the Internet to foment social unrest and influence election results in Europe and the United States.
Given this context, I am not keen on OneWeb sharing technology with Russia before or after a deal is closed.











Monday, November 26, 2018

A cool simulation of SpaceX's revised satellite broadband plan

On November 15, the FCC approved a revision to the plan for Starlink, SpaceX's forthcoming broadband satellite service. The new plan reduces the number of satellites from 4,525 to 4,409 and lowers the altitude of the phase-1 satellites from 1,100 to 550 km.

Revised broadband satellite plan (Sources 12 and 3
Mark Handley, a professor at University College London, has created videos based on simulations he ran of both the original and revised phase 1 plans. You should watch the videos, but it will help to first look over a few images from his simulation of the new plan.

Starlink phase-1 is shown below. Handley assumed a phase offset between adjacent planes that would minimize the chance of a collision. The minimum crossing distance between satellites is a little over 90 km -- an improvement over the previous plan.

Starlink phase 1

Looking up, there are several satellites in view at any one time (but fewer than there would have been with the previous 1,100 km plan):

View from the ground

Radio transmission will be used for links between the satellite constellation and the ground and laser links will be used between satellites. Handley assumed that each satellite links to the one before and after it in its plane and to the second closest in the adjacent planes in order to improve east-west routes. In his simulation of the first plan, he assumed a fifth laser linked to a satellite in the crossing plane, but that would be more difficult at this fast-moving lower altitude. It will be interesting to see how SpaceX uses the inter-satellite links.

Each satellite is linked to four others, forming a mesh network.

A simulated route between San Francisco and London is shown below. Note that latency is about half that of current, terrestrial routes. (The latency advantage of satellite routes over terrestrial is generally greater for long distances).

San Francisco-London -- twelve hops in space

Nine parallel routes between New York and London are shown below -- eight have lower latency than the current, terrestrial route.

Eight New York to London routes are faster than a terrestrial route

The following image shows all three phases -- Handley is not sure if the other two will interact with the phase 1 satellites.

Red: phase 1, white: phase 2, Blue, phase 3

Only Elon Musk knows what motivated this revision in SpaceX's Starlink plan, but we can speculate. They have had two test satellites in orbit at around 550 km altitude for some time, and that experience must have informed the decision. Musk reported that the links between the test satellites and the ground were performing well, but did not mention the laser links between satellites. Perhaps they feared difficulty with the longer laser links at 1,100 km or some other engineering problem. (Musk recently fired several managers in an effort to keep the project on schedule).

Because SpaceX and others are planning to launch thousands of new low-Earth orbit satellites, the FCC is focusing their attention on debris mitigation. The chance of a collision with the new plan is reduced since the minimum distance between satellites increases from over 40 to slightly over 90 km and.

The lower-altitude satellites may require more fuel to overcome a slight increase in atmospheric drag, but they will be easier to de-orbit when their life (around 5 years) is up, further reducing the likelihood of a collision.

Starlink's primary goal is affordable global broadband service to individuals and organizations, but Musk has also predicted that they would carry half of the global long-distance traffic. Handley's simulations predict roughly 2X latency improvement over current terrestrial cables on long routes and he points out that low-latency long-distance service will appeal to financial companies with offices in Europe, North America, and Asia. If capacity turns out to be sufficient, I imagine Starlink will also attract tier 1 ISPs and large companies like Google.

Finally, SpaceX also has approval for a constellation of 7,518 very low-Earth orbit satellites operating at altitudes from 335km to 346km. These will not be launched until SpaceX has satellite engineering, regulatory and market experience and time to assess potential cooperation/competition from 5G terrestrial networks.

I recommend watching the videos of both of Handley's simulations because the narratives are different and both are inciteful. You should also read his paper Delay is Not an Option: Low Latency Routing in Space.

Friday, November 16, 2018

An in-depth interview of OneWeb Executive Chairman Greg Wyler

OneWeb is building a large constellation of low-Earth orbit (LEO) Internet-service satellites and Via Satellite has published the "definitive 2018" interview of OneWeb founder and Executive Chairman Greg Wyler. The following are some of the quotes that caught my eye:
  • The system has been designed. The satellites have been tested. They are going through final stages of testing now before the launches begin. The satellites have actually performed better than expected in many ways, especially with their Radio Frequency (RF) performance which is really positive.
  • I think we will have customers up and running in 2020.
  • Whether [our satellites] are $500 thousand (the estimate in 2015) or $1 million is virtually irrelevant because what they are not is $50 million, and that is where it started.
  • The initial customers will be in the mobility and emergency services markets (paraphrase).
  • Aviation is a big [market] for us.
  • Why not let Sprint, DT, roam onto the plane? You can give the customers 4G/5G on the same devices they are used to using in their car, at the gate, or in other places.
  • The plane itself can become a Local-Area Network (LAN) party! I have been in aviation my whole life so this is always something I have been interested in.
  • OneWeb with its first constellation will be able to make a big impact on health centers and schools.
  • I would like to keep [the number of satellites up in five years time] below 1,500.
The tone of the interview was positive, but the early emphasis on emergency and mobile services (where they will have competion from other, relatively focused LEO satellite companies like Telesat and Leosat) makes me wonder whether their goal of eliminating the digital divide by 2027 might be slipping.

If I could have asked one question, it would have been about the objection to OneWeb that has been raised by the Russian Federal Security Service (FSB). If the FSB succeeds in stopping OneWeb in Russia, they will lose access to a potential market. Furthermore, it would jeopardize their contract for 21 launches with the Soviet space agency Roscosmos and perhaps cost and delay the project.

This has been a quick summary of a long interview -- you should check out the full interview.


Friday, October 26, 2018

Obstacles in OneWeb's negotiations with Russia

This case illustrates the fact that political, security, and financial negotiations may be as difficult as designing satellites and rockets for a would-be global Internet service provider.

OneWeb is investing billions of dollars in a constellation of low-Earth orbit (LEO) Internet-service satellites.

In 2015 they placed launch orders for 21 Russian-made Soyuz rockets.

In 2017, they formed a joint venture with Russian LEO satellite operator Gonets to develop the project in Russia. At that time, Gonets was a subsidiary of Roscosmos, the Russian State Corporation overseeing and implementing the Russian space industry. OneWeb had a 60% interest in the joint venture.

This week Reuters reported that OneWeb is relinquishing its majority stake in the venture -- Gonets intends to increase its stake to 51 percent.

I wonder why.

Speaking at a conference in Moscow, Federal Security Service (FSB) official Vladimir Sadovnikov objected to the project for security reasons. He feared that “Some of Russia’s regions would become totally dependent on a foreign satellite service” and added that Moscow had not received any conclusive evidence that OneWeb’s satellites would not be used for intelligence gathering.

(He also revealed his ignorance by apparently not understanding the difference between Iridium and OneWeb).

I wonder if the security concern is genuine -- OneWeb has decided to forgo inter-satellite links in favor of routing all traffic through a system of 40 terrestrial gateways, allowing a nation to know the path of traffic into and out of their territory. Are they concerned about the possible detection of sources of trolling and hacking?

Sadovnikov added a political dimension saying Russia favored setting up a similar network partnering with India, China and countries which he described as non-aggressive and China has pitched a 1,000 LEO satellite project to Russia.

An unnamed source at the FSB also mentioned politics, saying “OneWeb is an important project for Roscosmos and Russia’s space industry, but national security issues come first. There are many doubts regarding that project, especially because of the sanctions against us."

Spectrum is another stumbling block. OneWeb's request to receive a frequency band in Russia was refused and a source at the Ministry for Digital Development and Communications said they would be given permission after legal issues regarding the joint venture were completed. Given Russia's reputation, one can't help wondering whether the hangup has something to do with payoffs.

Another possibility is convoluted economic infighting within Russia. Gonets' Wikipedia page says it began as a Russian Federal Space Agency program, but in 1996 it was privatized and operated by Gonets SatCom, which was controlled by ISS Reshetnev. In 2017 Roscosmos acquired 80% of Gonets from ISS Reshetnev. Wikipedia is not a definitive source and I know nothing of the history of these organizations, but this sounds like it could be the kind of oligarchy-creation manipulation that occurs when state property is privatized. (The ownership of Cuban ISP Etecsa raises similar questions).

Perhaps there were management problems. Initially, launches of production satellites were planned to begin last May, then the date slipped to first quarter 2018. The current schedule calls for the launch of test satellites on February 7, 2019.

Regardless of the motivation for restructuring the OneWeb/Gonets venture, there is a mismatch in the aspirations of a global ISP and nationalistic governments. This case illustrates the fact that political, security, and financial negotiations may be as difficult as designing satellites and rockets for a would-be global ISP.

For background on OneWeb and other low-Earth orbit satellite Internet service projects, click here.

Gonets home page, 8/10/2018. It was removed earlier this week.

The Russian home page has also been removed. Last archived copy 4/10/18.

=====
Update 11/15/2018

Kommersant, a prominent Russian newspaper, reports that Russia’s Federal Security Service says OneWeb’s low-Earth orbit Internet-service constellation would pose a national security threat and wants it banned. If OneWeb service were banned in Russia it would mean the loss of a market for them and their Russian partner Gonets and would also jeopardize Russian space agency Roskosmos’ contract to launch OneWeb satellites. At the very least, that would probably delay OneWeb's plans.

The article says the issue has been raised "at the level of the President" in a Security Council meeting and a decision will be made no earlier than December. The English-language edition of Kommersant has a shortened version of this article. (I have based this note on a Google translation of the original article).

Thursday, October 11, 2018

Google -- kill Google Plus but save G+ Communities.

Last March, Google discovered a bug that exposed around 500,000 Google Plus profiles. Only static, optional profile fields like name, email address, occupation, gender, and age were exposed -- no other information like post content, Community memberships, viewing history, etc.

They discovered and immediately patched the bug last March and spent six months investigating it before going public yesterday. They say their investigation found "no evidence that any developer was aware of the bug or abusing it" and "no evidence that any profile data was misused." In spite of their assurance that no harm was done, Google stock dropped 2.6% last Thursday morning, when the news may have leaked out, and it is down about 4% for the last five days.

In the same post as they announced the bug, they announced they would "sunsetting" the consumer (free) version of Google Plus (but not the paid, enterprise version). Google or any other company has the right to discontinue an unprofitable product or service and, as they point out in their post, the "social network" portion of Google Plus was a failure, but there is more to Google Plus than social networking.

I quit reading my Google Plus feed long ago -- it was filled with spam and fake news based on my (faked) political interests, but I have found Google Plus Communities to be valuable and useful. While Google could not compete with Facebook's social networking feed, the features and interface of their Communities are superior to Facebook Groups. (Even if you like Facebook Groups, there is no way to transfer the members and history of a Google Plus Community to a Facebook Group).

Google should save Google Plus communities. There is a precedent for such a move. When it was launched, Google Plus included a service called Hangouts on Air (HoA). HoA enables video "chats" among up to ten people. That is not unique, but the chats can optionally be broadcast online and archived on YouTube. HoA was and remains a unique, valuable service that I and many others use. In 2016, Google removed HoA from Google Plus and integrated it into YouTube so it will not be affected by the elimination of Google Plus.

The decision to zap Google Plus provides a good example of the danger of dependency. There is an Internet saying -- "do what you do best and link to the rest." That makes sense and it has facilitated the rapid proliferation of Internet-based services, but it also leaves one vulnerable. The cost of using a service you depend upon may rise or one day or, like Google Plus, it may disappear.

I use HoA and Communities in my teaching and other professional work. Google saved me when they moved HoA out of Google Plus and I hope they do the same with Communities.

=====
Update 12/4/2018

Edward Morbius recently analyzed a random sample of 12,000 of the 7,974,281 Google Plus Communities. The size of the 10,358 Communities with visible membership ranged from 1-271,297 and the distribution was highly skewed -- the mean was 128.6, but the median was only 2.

Some Communities are worth saving, others are not. Communities with only one or two members are not worth saving, nor are communities with no recent activity (though they might have some archival value). Some topics should also be eliminated. For example, Morbius found many porn Communities in his sample and a Google algorithm should be able to flag and eliminate pornography. Spam users should also be detectable.

The largest Community in Morbius' sample had only 271,297 members, but there are much larger Communities. (I am a member of three Communities with over 500,000 members). Large Communities with non-spam traffic should be preserved. Small, active Communities are also valuable. For example, I create Communities for my classes. They are short-lived -- one semester -- and active relative to the number of members. They are quite useful to me and I do not know of a suitable alternative and I imagine the same is true for other finite-length projects.

Saturday, September 22, 2018

The impact of Internet-enabled gerrymandering: North Carolina

The Supreme Court sidestepped the North Carolina gerrymandering case and the 2018 congressional election will be biased there and in several other states. What is Judge Kavanaugh's position on the constitutionality of gerrymandering?

Charles and David Koch
Every ten years, after the census, states redraw the borders of their congressional districts. In most, but not all, states politicians -- the state legislature and governor -- control that process. The number of congressional districts in each state is a function of its population. For example, California, the most populous state, has 53 representatives in Congress while the seven smallest states have only one each.

Partisan gerrymandering is drawing district boundaries so as to give the controlling party in a state an unfair advantage. A gerrymandered map packs opposition party voters into one or a few districts so the controlling party has a small but safe advantage in the remaining districts. Gerrymandering is nearly as old as our country, but it was imprecise and difficult to do before the rise of the Internet. By 2010, the Democrat and Republican parties had compiled extensive voter databases, which mapping software could use to generate district maps favoring their party.

After the 2010 census, the Koch brothers and other wealthy Republican donors realized that if they controlled state governments, they could control redistricting and therefore the US House of Representatives so they increased their support of state elections and at the same time, the Democratic National Committee moved in the opposite direction -- focusing on the national election.

The Republican strategy worked. In the 2010 election cycle Democrats took five governorships from the Republicans while Republicans took 11 governorships from the Democrats and Republicans netted 721 state legislature seats, giving them control of both houses in 25 states -- up from 14 before the election.

Before and after 2010 gerrymandering (red=Republican control of both
houses, blue=Democrat, yellow=split). Nebraska is non-partisan. (Source)

The Republicans drew gerrymandered maps in key states and it paid off. In the 2012 election, 1.4 million more Americans voted for Democrats than Republicans for Congress, but Republicans won a 33-seat majority in Congress. A 2017 Brennan Center for Justice study concluded that "Republicans derive a net benefit of at least 16-17 congressional seats in the current Congress from partisan bias."

The Brennan Center uses three different tests for significant partisan bias in a state and they found consistent high bias in 8 states, four of which were found to be biased using all three of their measures.


There have been a number of court challenges in these states, most recently in North Carolina where a panel of three federal judges ruled last month that the state's congressional map is an unconstitutional partisan gerrymander that favors Republicans. It is easy to see how they reached that conclusion.

North Carolina voter registration numbers from 2010, before gerrymandering, to 2018 are shown below.

North Carolina voter registration (source)

Democrats outnumber Republicans in each year, but what about Libertarians and unaffiliated voters? If we adjust, by assuming that all of the Libertarian and half of the unaffiliated voters favor Republicans over Democrats, we get the following breakdown:

Adjusted registration percents

Before gerrymandering, there were seven North Carolina Democrats in the 111th Congress and six Republicans. That dropped to four Democrats after gerrymandering and then to three.

Note further that the Republicans have not tried to hide their bias. David Lewis, senior chair of the North Carolina House Redistricting Committee, and Robert Rucho, senior chair of the Senate Redistricting Committee, who were responsible for developing the redistricting plan, engaged a well-known Republican redistricting consultant Thomas Hofeller to draw the map.

Dr. Hofeller testified that Lewis and Rucho instructed him “to create as many districts as possible in which GOP candidates would be able to successfully compete for office” and he complied, seeking “to minimize the number of districts in which Democrats would have an opportunity to elect a Democratic candidate.” Lewis also confirmed their intent in the following television interview:

 Clip1 Clip2                                                    Representative David Lewis

Using common sense and tests like those employed by the Brennan Center, it is easy to identify gerrymandering after the fact, but harder to generate agreed-upon impartial districts when new census data is gathered. There are guidelines like having about the same number of people in each district, making districts contiguous and compact and recognizing political subdivisions like city and county boundaries, but these are not hard and fast criteria. There is also an inherent tension between the desire to avoid gerrymandering and compliance with the Voting Rights Act that prohibits racial discrimination in voting -- districts must be drawn so as to give representation to minorities.

People move around, so it is reasonable to redistrict every ten years and there is no way to get consensus on a redistricting algorithm, but it is clear that partisan politicians should not have redistricting authority. The Brennan Center study found one political party had sole control (legislature and governor) in each of the seven states with high levels of partisan bias while maps drawn by commissions, courts, and split-control state governments exhibited much lower levels of bias, and none had high levels of bias persisting across all three of the elections since 2010. You can read their recommendations here.

The North Carolina gerrymander was struck down by a Federal Court, but that was appealed to the Supreme Court. The lack of precise gerrymandering criteria led the Supreme Court to sidestep the case and send it back to North Carolina, where it was again found unfair, but there was not enough time to redraw the map before the election. The gerrymandered map will be used in the 2018 election. (The Pennsylvania State Supreme Court redrew a gerrymandered map this year).

If partisan gerrymandering persists, cases like that of North Carolina will continue to be appealed to the Supreme Court and they will eventually make a precedent-setting ruling. Today's Supreme Court is clearly politicized and a decision on gerrymandering would be more central to our democracy than rulings on important issues like DACA or abortion. What is Judge Kavanaugh's position on the constitutionality of gerrymandering?

For more on Internet-enabled gerrymandering see these posts and visit The Gerrymandering Project.

Tuesday, September 04, 2018

Might CubeSats provide broadband Internet connectivity one day?

It's hard to believe that the capability of the satellites companies like OneWeb and SpaceX are contemplating might one day fit in a CubeSat, but think about the phone in your pocket.

In November, 2016, SpaceX filed a request for approval to launch 4,425 Internet-service satellites using the Ku and Ka frequency bands. The satellites were expected to measure 4 x 1.8 x 1.2 meters. In February, 2018 SpaceX launched two Internet-service test satellites -- TinTin A and B -- that measured only 1.1 x .7 x .7 meters.

Why the size difference?

Maybe some functions were omitted from the test satellites -- for example, they may not have included inter-satellite laser communication capability -- but technical progress also steadily reduces the size of electronic devices.

Might the capability TinTin A or B one day be packaged in a CubeSat?

A CubeSat is a small satellite composed of one or more 10 x 10 x 10-centimeter cubes (units) that weigh under 1.33 kg each. For example, the 3-unit (3U) CubeSat shown below would be 10 x 10 x 30 centimeters (plus a little more for the frame holding the cubes) and weigh under 3.99 kg.

Assembling a 3U CubeSat

CubeSats were initially developed to support low-cost access to space for university research, but today they are being used in both commercial and research projects and many companies are manufacturing CubeSats and components.

Several startup companies are working on narrowband communication applications. Let's look at two examples.

SAS equatorial orbits (source)
Skyandspace (SAS) is off to an early start. They have three 3U CubeSats in an equatorial orbit and they have demonstrated instant messageing, voice calls, financial transactions and integration with public switched telephone network.

They plan to begin launching a 200-satellite constellation next year and to be fully operational in 2020. When complete, the constellation will serve the +/- 15-degree latitude region and provide personal voice calls and messages and connectivity for machine-machine communication and "Internet of things" applications.

SAS CEO Meir Moalem estimates that the "full constellation of the 200 nanosatellites will cost somewhere between $120 and $160 million,” which, he said includes "the production, launch, and operation of the constellation." They expect that that relatively low cost will enable them to replace 25% of the constellation each year, enabling them to constantly upgrade their technology. (APP Company Research used a figure of $150 million in an independent forecast).

SAS hopes to eventually provide global narrowband coverage with 1,000 satellites, but they are not talking about broadband service. On the other hand, "Internet in space" is the long-term goal of Kepler Communications.

Kepler KIPP (source)
Just over two years after the company was founded and a year after being funded, Kepler had KIPP, their first operational satellite, in orbit -- a testimony to CubeSat cost and development time.

KIPP, a KU-band, 3U CubeSat, is in polar orbit and providing high-speed, global store-and-forward service through their gateways in Inuvik and Svalbard. They are serving customers with latency-tolerant applications like bulk transfer of scientific and video data. (This is reminiscent of the first satellite Internet project I know of -- VitaSat -- which provided email and other asynchronous services in developing nations in the mid-1990s).

The next launches will be CASE, a slightly upgraded version of KIPP, which will add to their store-and-forward capacity and TARS, a redesigned 6U CubeSat that will be used for both store-and-forward and Internet of things applications. TARS will be the final service-demonstration satellite prior the early 2020 launch of the first 10 satellites of their 140 satellite constellation.

Both SAS and Kepler are small startups. SAS has issued stock and Kepler is venture funded. Kepler has also received In-Orbit Demonstration Mission funds from Satellite Applications Catapult, a non-profit company funded by Innovate UK which in turn is funded by UK Research and Innovation. (This sounds like an interesting public-private funding chain).

Like OneWeb (and unlike SpaceX), both work with partners in design and production of satellites, antennas, radios, etc. and both sound like fast-moving, innovative companies with a sense of purpose -- reminiscent of the early days of both personal computing and the Internet. The "careers" page of the Kepler Web site captures this feeling well. It offers the opportunity to "join the team that's building the Internet in space," recognizes that "a good cultural fit can oftentimes be more important than technical competency when building a company" and lists "indoor bike storage" as one of the perks of employment. I'm ready to sign up!

It's hard to believe that the capability of the satellites companies like OneWeb and SpaceX are contemplating might one day fit in a CubeSat, but think about the phone in your pocket. Also, as the following time-lapse video (1:54) shows, today's CubeSats are hand built -- what would mass-produced CubeSats cost and how many could a BFR launch at once?

Tuesday, July 24, 2018

The impact of rising sea level on Internet infrastructure

A recent study predicts that rising sea level might result in as much as 4,067 miles of fiber conduit being under water and 1,101 nodes (data centers, Internet exchanges, cable landing points, etc.) surrounded by water in U. S. coastal cities in 15 years.

Paul Barford, professor of computer science at the University of Wisconsin, and his colleagues have been compiling data on the physical Internet and making it available to the research community at the Internet Atlas Web portal since 2011. The portal includes an interactive visualization tool that can be used to create maps of Internet infrastructure in order to inform the public and guide network architects and policymakers.

Barford and his colleagues have combined their infrastructure data with sea level incursion projections from the National Oceanic and Atmospheric Administration (NOAA) in order to estimate expected damage to U. S. Internet infrastructure due to sea level rise and, as you see below, the NOAA data shows that sea-level has increased relatively rapidly off the U. S. coasts in recent years.


That is history, but NOAA also models future sea level rise and you can see predictions for specific locations using their Sea Level Rise Viewer (SLRV), a cool visualization tool. The SLRV predicts sea level rise through the year 2100 and computes estimates of sea level rise for five different scenarios at a selected location. For example, predictions for Miami, Florida for the year 2030 are: Intermediate Low scenario: 0.49ft, Intermediate: 0.75ft, Intermediate High: 0.98ft, High: 1.21ft and Extreme: 1.35ft. (The assumptions underlying the scenarios are spelled out here)

For this study, Barford and his colleagues used NOAA's Extreme scenario estimate of a global mean sea level rise of one foot per 15 years through 2100 and concluded that in the U. S. 4,067 miles of fiber conduit will be under water and 1,101 nodes (e.g., points of presence, Internet exchanges, cable landing points, etc.) will be surrounded by water in the next 15 years.

National outage due to hurricane
Sandy (source)
The study concluded that Miami, New York, and Seattle will be the hardest hit U. S. cities, but don't think you are home free if you live in Kansas -- the Internet is global and we are all affected by outages.

They also noted that since much Internet infrastructure is located near the coast in large cities, much of the future damage will occur fairly soon. Even if we assume the Intermediate Low scenario, Miami sea-level will have risen by a foot by 2015.

Los Angeles in 15 years
You can use the SLRV to check the situation in your local area using the scenario that best fits it. For example, this is what the coastal region of my city, Los Angeles, is projected to look like in 15 years using the Extreme scenario. The blue areas (on a green background) show expected sea level rise inundation and the dark green lines are fiber conduits.

Don't forget that Barford's analysis assumes the most aggressive of NOAA's scenarios and at the historical rate of sea-level rise we would predict less than 2 inches in the next 15 years.

On the other hand, Barford used the mean global extreme scenario estimate and North American cities have historically seen above average sea level increases.

Furthermore, he does not consider storm damage and this report shows that when hurricane Sandy hit the east coast in 2012, outages in New York and New Jersey led to a doubling of the national Internet outage rate. (You can read more on the impact of Hurricane Sandy on the Internet here).

Hurricane Katrina also damaged the Internet when it struck New Orleans in 2005. You can get a feeling for what it was like by listening to Doc Searle's interview of Sigmund Solares, then CEO of the DirectNIC data center. Solares describes the flooding and the extreme effort required to keep DirectNIC online.

(Note that the posts on Sandy and Katrina are at the Internet Archive since the original copies were no longer online).

Some people believe global warming is God's will -- it's good that He also endowed us with free will.


Thursday, July 19, 2018

Trump's "would/wouldn't" comment on Russian meddling in context

Whether he meant to say "would" or "wouldn't," trump clumsily tried to dodge the question of Russian interference in our election by falsely attacking the FBI and taking a shot at Hillary Clinton and a "Pakistani gentleman."

The news has focused on Trump's response when Associated Press reporter Jonathan Lemire asked whether he believed Putin or "every U. S. intelligence agency" on the question of Russian interference in the 2016 election. Trump and the media have focused on this quote:

I will say this: I don’t see any reason why it would be, but I really do want to see the server.
The next day, Trump claimed to have misspoken, that he meant to say "wouldn't" instead of "would," but let's look at that sentence in context. It turns out he tried to artfully dodge the question by attacking the FBI. Here is his answer:
So let me just say that we have two thoughts. You have groups that are wondering why the FBI never took the server. Why haven’t they taken the server? Why was the FBI told to leave the office of the Democratic National Committee? I’ve been wondering that. I’ve been asking that for months and months, and I’ve been tweeting it out and calling it out on social media. Where is the server? I want to know, where is the server? And what is the server saying?

With that being said, all I can do is ask the question. My people came to me — Dan Coats came to me and some others — they said they think it’s Russia. I have President Putin; he just said it’s not Russia.

I will say this: I don’t see any reason why it would be, but I really do want to see the server. But I have — I have confidence in both parties. I really believe that this will probably go on for a while, but I don’t think it can go on without finding out what happened to the server. What happened to the servers of the Pakistani gentleman that worked on the DNC? Where are those servers? They’re missing. Where are they? What happened to Hillary Clinton’s emails? Thirty-three thousand emails gone — just gone. I think, in Russia, they wouldn’t be gone so easily. I think it’s a disgrace that we can’t get Hillary Clinton’s 33,000 emails. (Source).
He used the word "server" nine times and managed to attack Hillary Clinton and a "Pakistani gentleman" along with the FBI.

We know what he was getting at when he mentioned Hillary Clinton and "the Pakistani gentleman that worked on the DNC," but what about the server?

I listened to the press conference live and was struck by his rambling reference to servers. I didn't know what servers he was talking about and also wondered whether he knew what a server is. (It turns out he repeated the word nine times).

I did a quick search and it turns out that the fact-checking site Politifact has concluded that "Donald Trump's 'missing' server comments get all of the details wrong" and the Pakistani gentleman is Imran Awan, who emigrated from Pakistan as a youth, has been a U.S. citizen since 2004 and did not work for the DNC.

Tuesday, July 10, 2018

Telecommunication in ancient Greece

The Greek historian Polybius (c. 200 – c. 118 BC) documented two methods for communication at a distance -- telecommunication.

The first was an analog hydraulic system. The terminals consisted of identical vessels filled with water. The sender signaled the receiver with a torch when he was ready to transmit and they both let water out of their vessels until the end of transmission torch was raised. If the amount of water drained from both vessels was the same, the height of the water in the receiver designated the message.


This method was imprecise and could only transmit one of a number of pre-defined messages. The second, digital method could transmit any written message. It used a 5-bit representation of characters to send messages using torches. Messages were transmitted one character at a time by raising torches in two of five positions.


The operators had identical tablets showing the letters and the torch positions (1-5) designated the column and row of the letter being sent. If this tablet were used:


the message 22, 51, 13, 13, 45 would spell out ΗΕΛΛΩ.

Note that messages could be encrypted by changing the tablets.

(If you find this interesting, check out Claude Chappe's French telecommunication system).

Wednesday, June 06, 2018

How about opening a YouTube video production space in Havana?

YouTube has video production spaces in Los Angeles, London, Tokyo, New York, São Paulo, Berlin, Paris, Mumbai, and Toronto -- how about opening one in Havana?

In 2015 I wrote that Cuba was well positioned to produce Spanish-language entertainment and education content and suggested that they open a YouTube production space in Havana.

The idea of opening a production space in Havana makes even more sense today than it did in 2015.

Google Global Cache (GGC) servers are now installed and operating on the Island. That means Google content can be viewed and uploaded faster than in 2015 and the result is that YouTube has made substantial gains since GCC went online last April:

YouTube gains since GCC went online (source)

There is also a growing, enthusiastic community of young Cuban YouTubers, several of whom are profiled in this YucaByte article and you can "meet" a few others in this short (5:36) video from Periodismo de Barrio:


(The YucaByte article also contains a short video sampler showing enthusiastic YouTubers).

Not exactly being a millennial myself, I am not likely to become a follower of these youthful YouTubers, but they are inheritors of a rich history of Cuban music, cinema and education.

Google executives and Senator Jeff Flake just met with Cuban president Diaz-Canel and former Google Chief Executive Eric Schmidt who said “We already have an agreement signed that allows easier access to data. We would like to do more.”

How about opening a YouTube video production space in Havana?

Tuesday, June 05, 2018

SpaceX Starlink test -- good news but unanswered questions

Last week, Elon Musk was asked on Twitter how the Starlink tests were going and he replied that the two test satellites, TinTin A and B, are connecting at "high bandwidth" with 25 ms latency.

That's good news but it leaves a lot unanswered. For example, he did not mention the speed and reliability of the phased-array handoffs between the satellites and ground terminals and he said nothing about tests of the inter-satellite laser links, which OneWeb has abandoned.

While we have experience with radio links between satellites and the ground, inter-satellite laser links are new so I'm more curious about those tests. What sorts of speeds and latencies are they seeing on transmissions between TinTin A and B and how well are they doing at creating and maintaining links between the satellites? Fast inter-satellite switching and transmission speed are critical to overall performance of the constellation grid, particularly on long-distance links.

Click here for a survey and updated progress report on SpaceX Starlink and other potential LEO-satellite based Internet service providers.

Friday, May 25, 2018

SpaceX President and COO Gwynne Shotwell on synergy among Musk companies and Starlink profit

Source
SpaceX President and COO Gwynne Shotwell gave a recent interview in which she said that SpaceX is profitable, but she predicts a much larger market for the Starlink Internet service. (As we see here, a January 2017 Wall Street Journal article made the same point).

Shotwell also spoke of synergies among Elon Musk's companies: Tesla cars will be online via the Starlink Internet service; Tesla battery technology has been leveraged for the Falcon 9 rocket and Dragon spacecraft and Boring Company technology will be used in housing construction on Mars. They have also learned manufacturing techniques from Tesla and will be able to produce one rocket engine per day and two complete Falcon 9 rockets per month.

She also said they remain on schedule to take people to Mars in 2024, and, when asked about Elon Musk, she said he spends about half his time on SpaceX and half on Tesla and that he is an inspirational leader.

Click here for a survey and updated progress report on SpaceX Starlink and other potential LEO-satellite based Internet service providers.

Wednesday, May 23, 2018

Telesat begins testing low-Earth orbit satellite Internet service

SpaceX and OneWeb get a lot of publicity and have ambitious plans, but Telesat is the first low-Earth orbit (LEO) satellite Internet service provider to begin testing with potential resellers.

Last January, Telesat launched a demonstration satellite and it is now ready for testing. Maritime connectivity provider OmniAccess and Australian ISP Optus had committed to testing the system previously and this week they were joined by in-flight entertainment company Global Eagle Entertainment.

Global Eagle CEO Josh Marks said he was persuaded to collaborate with Telesat by their planned coverage over oceans, polar regions and high-latitude routes and their "open architecture" business model. In addition to testing, they "will collaborate with Telesat on both the technology and commercial model for their new LEO platform.”

OneWeb and several airlines have formed the Seamless Air Alliance, which is developing standards for in-flight Internet connectivity through LEO satellites. I wonder whether Telesat and Global Eagle will join the alliance or go their own way.



Click here for a survey and updated progress report on Telesat and other potential LEO-satellite based Internet service providers.

Monday, May 21, 2018

Elon Musk tells what to expect from the Block 5 Falcon 9 rocket

The ability to launch 30 Falcon 9s per year at a cost of $5-6 million per launch, would be a big plus for SpaceX's Starlink Internet service.

On May 11, SpaceX launched a Bangladeshi satellite using their Falcon 9, Block 5 rocket. This was the first production flight for the Block 5. The day before the launch, Elon Musk participated in a call with reporters and the following are some of the points he made. (You can read more analysis and read a full transcript of the call here)

SpaceX accounted for over half of US launches in
2017 and expects to double their launch rate.
In 2017, SpaceX had 18 successful launches and Musk stated that they were on track to double their launch rate this year, implying a rate of 3 launches per month. He said that "if things go well, which is a caveat, then SpaceX will launch more rockets than any other country in 2018."

There will not be a Block 6. Musk said that after 8 years of upgrades, the Block 5 will be the last major version of the Falcon 9 before their next rocket, the BFR.

Musk expects the Block 5 "to be a mainstay of SpaceX business," and there will be 300 or more Block 5 flights before it is retired in favor of the BFR.

The Block 5 is designed for rapid-turnaround reusability. It is "designed to do 10 or more flights with no refurbishment between each flight — or at least not scheduled refurbishment between each flight. The only thing that needs to change is you reload propellant and fly again." He also said that "the Block 5 boosters are capable of on the order of at least 100 flights before being retired."

Musk has set a goal of demonstrating "two orbital launches of the same Block 5 vehicle within 24 hours, no later than next year."

The Block 5 was designed "to be the most reliable rocket ever built." They have exceeded all of NASA's human-rating requirements and have met "all of the Air Force requirements for extreme reliability."

Reliable reusability will cut cost dramatically. Musk broke down launch cost as follows: booster about 60 percent, upper stage 20 percent, fairing 10% and the launch cost 10%. If they are able to reuse all three rocket elements, they would be able to "reduce the cost for launch by an order of magnitude ... to $5-6 million per launch." Musk pointed out that getting to this point had taken "16 years of extreme effort" (and a lot of learning from failures).

The ability to launch 30 Falcon 9s per year at a cost of $5-6 million per launch, would be a big plus for SpaceX's Starlink Internet service.