Reports of a Significant Gap are greatly exaggerated…

(NOTE: This is the second part in our series on Crown Castle’s Bad Plan for Palos Verdes, part 1 is here.  The post gets into technical details that may not be for everyone, but we hope readers will stick through it and apologies for the length.  As we’ve seen before, there’s plenty of troubling, unanswered questions that need an explanation.  It’s important to understand exactly what is taking place when exercising our right to regulate highly intrusive cell towers.)

The Meaning of Effective Prohibition

prop map example

A propagation map used in significant gap analysis

Under state law, cities in California can deny cell towers in the public Right of Way on aesthetic grounds (PUC 7901, Sprint vs PVE).  However, if a site is required to prevent an “effective prohibition” in service, then a city may be required to approve the tower despite this under the Telecommunications Act of 1996.  Federal law supersedes state law.

Two key concepts that have evolved in federal law are “significant gap” and “least intrusive means”.  If the applicant can prove:

  1. The site is required to close a significant gap in service AND
  2. The proposed method for doing so is the least intrusive means

THEN failure to approve the site is likely an effective prohibition of service.

However, it’s not an easy threshold to meet as evidenced by a string of lost lawsuits brought by the carriers.  If cities do their homework, and don’t do something stupid (such as denying sites based on perceived RF health effects), then they’ll be on firm ground.

But the carriers and their tower installers realize most cities don’t understand the law.   They can often be intimidated by a bunch of technical mumbo-jumbo, legal bluster, and copy and paste claims of entitlement.  We’ve certainly seen that tried in Palos Verdes.  The ridiculous sites proposed by Crown Castle were all claimed to be “least intrusive means” despite the fact they were shoddy designs, plopped right in peoples front yards, with a primary goal of minimizing costs.

Significant Gap in Rancho Palos Verdes?

So with that background, it’s not surprising that the carriers and tower installers claim virtually every proposed site is required to close a significant gap.  Fortunately, we have RF engineering expertise and can test Crown Castle’s claims to see how well they hold up.  (Spoiler alert: they don’t)

Here’s the “significant gap” narrative from the site justification for proposed site ASG 53 in Rancho Palos Verdes (highlight added, original here):

Crown Castle significant gap claim

In summary, Crown Castle claims the entirety of RPV is just one giant significant gap.  No joke, in the highlighted red box they state “AT&T has in effect little to no coverage for the balance of the city”.  Sounds pretty bad, and they submit a “drive map” as evidence.  (Side note: RPV’s wireless ordinance requires propagation maps, drive maps are not an acceptable substitute.)


Crwon Castle drive map RPV

Drive map submitted to Rancho Palos Verdes by Crown Caste showing alleged ATT LTE coverage.  Note red box added to plot, it will be important below. Source: RPV public records, original here

Wow, that looks like awful LTE coverage.  Crown Castle states a signal level of -85 dBm or greater is required for adequate coverage thus only the dark green, light green, and yellow streets have sufficient coverage.  All the streets shown in red, light blue, or dark blue are lower than -85 dBm (more negative) thus they supposedly have a significant gap.

Fortunately, this didn’t pass the sniff test and let’s just say this claim set off bells like a three-alarm fire.  Over on the right hand side it shows the percentages for each of the coverage levels.  Signal levels of -85 dBm or greater (more positive) totaled 9.5%.  Thus Crown Castle claims 90.5% of the streets they drove have a significant gap in service.  Does anyone really believe this?  Cell coverage in Palos Verdes is not great, but it’s not that bad.  Time to call out these guys.

A Different Story in Palos Verdes Estates

Back in October of 2015, Crown Castle submitted actual propagation maps to the city of Palos Verdes Estates.  Fortunately, these maps also covered large areas of RPV so we had something to compare the RPV drive map shown above against.

Crwon Castle propagation map PVE

Propagation map submitted to Palos Verdes Estates depicting ATT LTE coverage.  Red box was added to plot to show area in common with the drive map submitted to Rancho Palos Verdes above. Source: PVE public records, original here

In stark contrast to the drive map submitted to RPV, this map shows most of the depicted area is -85 dBm or greater (green or yellow).  Also note that the areas lower that -85 dBm (light purple in this map) are primarily in the top left.  This area isn’t even in Rancho Palos Verdes but is actually in Palos Verdes Estates.  Unlike the map and site justification submitted to RPV, which claims the city is one big cell coverage wasteland, this map shows pretty good coverage.  What is going on?

There’s at least three highly questionable tactics at work here, so we’ll walk through them.

Tactic 1: Frequency Band Cherry-Picking

Alert readers may have noticed the first map depicted 1900 MHz LTE coverage while the second map depicts 700 MHz LTE coverage.  So why did Crown Castle change frequency bands when submitting maps between the two cities?  That is a really good question that gets at the heart of the first tactic.

Crown Castle’s cell towers will provide ATT service on four frequency bands in Palos Verdes, commonly referred to as the 700 MHz, 850 MHz, 1900 MHz, and 2100 MHz bands.  Blocks within these bands are all licensed to ATT.  Crown Castle says as much in the “power output and operating frequency” table submitted to RPV.  Yet Crown Castle only submitted the 1900 MHz drive maps to Rancho Palos Verdes and did not submit the other bands.  The omission is telling.

Here’s an example of the power output and frequency table, we’ve added the common frequency band designations below each column for clarity.  Also note the “technology” entry in the third row, we’ve reproduced it below and added the common designators.

Crwon Castle power and frequenciesCrown Castle’s power and operating frequency submission to Rancho Palos Verdes.  The same table has been submitted to RPV and PVE for all proposed sites with minor power variations depending upon the antenna selected. Text in red added for clarity. Source: RPV public records

There are two technologies shown, LTE (4G) and UMTS (3G and lower).  UMTS support is retained for older devices without LTE capability since LTE and UMTS use different modulation and aren’t compatible.  As these older devices disappear, bands are being transitioned from UMTS to LTE using a process known as refarming.

Now look closely at the 1900 MHz column shown above, Crown Castle states its use is UMTS (3G), not LTE (4G).  But the 1900 MHz drive map submitted for RPV doesn’t depict UMTS, it depicts LTE.  It says so right in the legend (see zoomed view below).  How can that be?

Crwon Castle drive map legend RPV 1

Zoomed view of submitted drive map legend shows coverage is LTE. Source: RPV public records

Now we get to the heart of this tactic.  The 1900 MHz band is a band in transition; it is in the process of being refarmed from UMTS (3G) service to LTE (4G) service.  This process takes time.  As Crown Castle’s table shows, the 700 MHz and 2100 MHz bands are the primary LTE services, 1900 MHz until recently was being used for UMTS (3G).  In summary, Crown Castle submitted only a single, partially-active, supplemental LTE frequency band as representing their LTE service in RPV.

All modern phones have multi-band capability allowing selection of the various available frequency bands.  LTE service in the city is the aggregate of all bands providing the service, not that of a cherry-picked, worst-case band.

Tactic 2: The Missing Macro Sites

On closer review, there is something else funny going on in the 1900 MHz maps submitted to Rancho Palos Verdes.  Take a look at the alleged “existing” coverage around the proposed site ASG53 submitted to RPV versus that submitted to PVE.  Notice anything missing?

Crown Castle missing macroZoomed in alleged “existing” coverage in vicinity of proposed site ASG53.  RPV submitted drive map on left, PVE propagation map on right.  Note the purple dot depicting an ATT macro site on the right. Legend from submitted map shown at top. Source: RPV and PVE public records

The PVE propagation map clearly shows there is an ATT high power macro site (purple dot) located only 850 feet away from the proposed site ASG53, yet Crown Castle claims coverage is non-existent.  The BS detector is going off full-blast again.

RPV’s wireless facility application process requires applicants to identify the “location of the proposed facility in relation to all existing and planned facilities maintained within the city by the applicant, operator, and owner, if different entities.”  In response to this requirement, Crown Castle submitted this map:

Crown Castle proposed and existing sites

Crown Castle submission to RPV purporting to show all existing and proposed locations. Source: RPV public records, original here

We’ll set aside the fact that the map is nearly unreadable, presented from an odd oblique angle, doesn’t identity any street names, and just focus on the missing macros.  Here’s the zoom in of the area around proposed site ASG53, the macro is completely absent.  In fact, all ATT macro’s are missing from the map as it only shows Crown Castle-owned small cells.

Crown Castle proposed and existing sites zoom

Zoomed in map showing the area around proposed site ASG53.  Existing macro site 850 feet away is conspicuously absent. Source: RPV public records

So perhaps this is just a clerical error.  Maybe the missing macro isn’t operational on 1900 MHz.  If the drive maps accurately show coverage, no harm, no foul, right?  That’s what we wanted to check so we set up for measurements near the site to determine if the macro site was, in fact, providing LTE service on the 1900 MHz band.

We used a spectrum analyzer with a directional antenna oriented toward the macro site.  Measurements were taken at four locations and show received isotropic power (RIP) after correcting for antenna gain and cable loss.  We don’t claim NIST-traceable accuracy, but the results are startling.  Looks like there might be a foul after all, we’d even call it a “technical” foul.

Missing macro field measurements

Results from field measurements conducted in the vicinity of the missing macro site.  Macro appears to be fully operational with AT&T LTE service on both the 700 MHz and 1900 MHz bands.

Unless there’s a pirate squatting on ATT’s federally-licensed frequency block, the submitted drive maps are in error.  There were clearly identifiable and strong LTE waveforms emanating from the macro site on both the 700 MHz and 1900 MHz bands within ATT’s frequencies.  Maybe the drive maps were conducted prior to 1900 MHz refarming of the macro site.  Perhaps.  However, the maps were submitted to RPV on 5/3/2016 and the measurements were taken less than three weeks later.  Regardless of how it happened, the maps don’t accurately represent current LTE service on the 1900 MHz band.

Tactic 3: Changing Service Metrics

If you refer back to the legend in the Crown Castle submitted drive map, you’ll see it says “LTE_Scan_RSRP_Sortedby_RSR(P)…”.  That provides another important clue as to what is actually being depicted in the drive map.

RSRP stands for Reference Signal Received Power and it’s one of several metrics used in evaluating LTE signal quality.  It’s along the lines of RSSI and RSCP often used on older 2G and 3G systems.  The propagation map submitted in Palos Verdes Estates almost certainly depicts RSSI based on the high signal levels and the alleged required signal level.  Those propagation maps were prepared by ATT, not Crown Castle.

In Crown Castle’s drive maps, they have switched to using RSRP but they failed to account for the fact that RSRP is measured and calculated over a much narrower bandwidth.  There’s a great explanation on the topic here.

RSSI measures the signal over the entire integrated bandwidth of the active transmitted LTE waveform, in the 700 Mhz case it is 9 MHz wide (10 MHz BW with guard bands).  In contrast RSRP represents reference signal power within a small fraction of the bandwidth, a 15 kHz sub-band known as a resource element.  The reference signal is present in every third resource element.  This results in RSRP levels being  roughly -23 dB (200 times) lower than RSSI levels.

RSRP to RSSI ratio = 10*log((15 kHz*3)/9.0 MHz) = -23.0 dB

Here’s a spectrum analyzer screen shot showing the difference.

LTE spectrum

Screen shot showing the overall LTE waveform (5 MHz here) vs the individual reference signal(red).  Note there is one reference signal for every three resource elements.

Due to this fact, good signal levels using the LTE RSRP metric are much lower than that seen in the past using older UMTS technology signal metrics.  This has caused confusion with early LTE phones showing low service (few or no bars) even when the LTE signal was relatively strong.  However, those paying attention realize using the old “required” signal levels of obsolete technology results in inaccurate apples to oranges comparisons.

This shift is clearly shown in recent court cases involving LTE.  In Verizon vs. Fairfax County (E.D. Virginia, 2015), Verizon’s RF expert stated required LTE RSRP for indoor service is -95 dBm.  In our opinion, even this value is high.  Assuming for the sake of argument that -95 dBm is the right value, that’s still 10 dB (10 times) lower than the -85 dBm Crown Castle says is required in Rancho Palos Verdes and 20 dB (100 times) lower than the -75 dBm they say is required in Palos Verdes Estates.  They have provided no explanation for the two different “required” signal levels in adjacent cities, and frankly we don’t expect one.

(BTW, Verizon lost this case on all counts, primarily because they continually contradicted themselves and had lousy documentation.  Sound familiar?).

Until recently, no one has been paying attention to this stuff.  We hope this post shows that we are paying attention, we are documenting everything, and don’t like what we are seeing.

Wrapping Up

If you made it all the way through this post, you have our sincere appreciation and admiration. It definitely gets in the weeds, but they are important weeds.

Maybe all these technical errors are just a big misunderstanding.  Maybe the mistaken photo simulations and distorted drawings were all just a big misunderstanding too.  Maybe the claims of Section 6409 applying to these sites when it actually doesn’t is also all a big misunderstanding.  Maybe we just have more misunderstandings than a typical old Three’s Company rerun.  Then again, maybe not.  We’d like an explanation how this stuff keeps happening.

Like this post?  Click here to go to our home page for more.