FCC Proposed New 6 GHz Wifi Spectrum

On October 2, the FCC proposed WiFi access to the 6 GHz region (5.925-7.125 GHz) in addition to the currently accessible 2.4 GHz and 5 GHz bands of frequencies. 

Currently most WiFi access points operate at either one of those 2.4 GHz of 5 GHz bands. Problems arise as more and more devices are connecting via WiFi. Due to longer wavelengths, 2.4 GHz band signals travel further but with the growing plethora of wireless devices, often suffer from congestion and interference. The 5 GHz band typically operates at higher speeds but does not travel as far due to the shorter wavelengths. 

Opening up the 6 GHz region will provide close to three times what is available in the 2.4 and 5 GHz regions - great for locations where lots of people are connecting at the same time (think college campuses, airport terminals, etc). 

The 6 Ghz frequency region is currently used for point-to-point microwave links and earth-to-space communications along with other data links and there will likely be some opposition. There is a current FCC public commenting period and there will be another vote once the commenting period is over. For details you can read the full FCC Notice of Proposed Rulemaking for ET Docket No. 18-295; GN Docket No. 17-183 linked here.

An Experience With An Intelligent Car

Yesterday I attended an excellent advisory board meeting for a National Science Foundation funded eBook project called E-MATE at Brookdale Community College in Lincroft, NJ. Mike and Kelly are doing some really cutting edge ground-breaking work in the development of electronic instructional materials and it was an excellent meeting. I need to do some writing here about the work they are doing. Today though – I want to write about cars.

Diane was away and I had the chance to drive her car (a 2014 Volvo XC70) back and forth to the meeting. We leased this car in December 2013 and she’s the primary driver.  Yesterday was my first opportunity to take this car solo (solo is the key word here) on a road trip of almost 500 miles. The car is loaded with just about every option including the technology package and I’ve been chomping at the bit to really give the technologies a test, especially after seeing one of the autonomous Google self-driving cars in downtown Mountain View a few weeks ago.

Volvo does not offer a self-driving package (yet) but my experience - it is pretty darn close to self-driving with the technology package that adds adaptive cruise control, automatic high beam control, frontal collision warning, automatic braking for frontal collision crash mitigation, a driver inattention monitor, blind-spot warning system, active xenon headlights, and lane-departure warning to an already incredibly safe and comfortable car.

Now - driving from Massachusetts to New Jersey on a weekday is always an experience – New York City cannot be avoided unless you want to add hours to the trip and that means bumper-to-bumper traffic, crazy drivers and lots of intense time behind the wheel.

I was so impressed with the car – stop and go for at least a couple of hours and no need to hit the brakes or the accelerator. It took some time to get used to – I had to “trust” the car but once I did – amazing! An alarm that goes off if the car starts to drift outside the lane (unless a directional has been used). Sensors that monitor and determine whether the driver is becoming tired and inattentive. Cameras that watch for speed limit signs and indicate when the speed limit has changed. A blind spot warning system that indicates a car is coming up from behind on either side. Sensors that monitor oncoming traffic and control high beams.

Does the car drive itself – no – not yet but it is pretty close. Did I push the technology? I don't think so. I let the car do what it is designed to do. What did I do? I pretty much steered and adjusted the cruise control up and down. I did not have to use the accelerator or brakes unless I wanted to on the highway, whether I was going 70 mph or in a stop and go traffic jam.

As an FYI Volvo in 2017 will start testing 100 "production-viable" autonomous self-driving cars in Sweden with real drivers like you and me. These test cars have 28 cameras, lasers, sensors, and radar units along with integrated computers and communications systems that make up the self-driving system.  How soon will we have the chance to purchase a self-driving car? Right now it is looking like 2020.

With my new position at the Center for Optics and Photonics Education and my past position at the Information and Communications Technologies Center, cars (and a lot of other devices) are really hitting a sweet tech spot for me. Infrared lasers, optical sensors, integrated GPS, radar and cameras collecting large amounts of data, onboard computers processing the data, communicating back to the car and driver and making intelligent "pretty-big-data" decisions. Super cool stuff and I’ll be writing over the summer about some of these individual technologies and how they work.

Now for me – it is back to my older Toyota product with none of the car sensor and intelligent technologies (it does have a back-up camera and Bluetooth). I have to remember when I’m driving my car all of the “intelligence” is up to the driver. Ohhhh Noooo :)

Storage Tiering

This is another topic I’ve been reading a lot about lately.  Storage tiering uses expensive faster access drives for frequently used data and slower less expensive  access drives for older archive type data. 

Typical fastest level tiers in a data center will use an optically connected fiber channel disk array, followed by Small Computer System Interface (SCSI) attached drives, followed by Serial Advanced Technology Attachment (SATA) drives followed sometime even by tape drives. Tape drives – I know – but yes they are still used - typically to take content completely offline and store.

Mechanical drives have been historically used for the fastest tier but as prices drop and operating systems include better support , we’re seeing a lot of much faster solid state memory devices (solid-state drives (SSDs) and I/O accelerator cards (sometimes referred to as solid-state accelerators [SSAs]) used for the faster level  tiers now.

What’s the difference between a SSD and an SSA? They both basically do the same thing – the only difference is the data interface. 

• SSD’s look like a mechanical hard drive and the server operating system uses standard BIOS calls to access. 
• SSA’s are a little different and use a PCI Express (PCIe) interface. The operating system must use a unique set of software drivers specific to the SSA device being used. 

Because SSA’s use drivers that have been tuned specific to the device, they tend to be a little faster. Both SSD and SSA devices can work together on the same server.

The Rise Of The HetNet

I’m starting to see the term HetNet used in my reading and even had a student ask me for details this morning. That said – I figured it would make a good blog post.

HetNet is short for Heterogeneous Network and is a term currently being used in the wireless world. Most of us are carrying around smartphones with a number of different antennas built in for access via different network technologies. The neat thing about HetNets is they are able to maintain mobile connectivity (no dropped connections) when switching between different wireless connection technologies. 

The three different connection technologies getting all the HetNet buzz right now are LTE, Carrier WiFi, and enterprise femtocells. LTE is 4G cellular service and Carrier WiFi is just WiFi service provided by a wireless carrier. Femtocells are small cell devices that are connected to an Internet broadband connection. 

So HetNets allow a user device to seamlessly switch from network type to network type– LTE to WiFi to femtocell and vice-versa back and forth without dropping a connection. Pretty neat.

Now, not too long ago, it was believed LTE would be the dominant mobile technology and there would not be a need to alternative type technologies like Carrier WiFi and femotocells. But think about it…… there is only a limited amount of spectrum and bandwidth so providers are looking for ways to lower the number of devices per cell. There are also advantages to having users as close as possible to the different types of base stations. 

As a result, we’re seeing providers like Verizon Wireless and AT&T use HetNets to improve the coverage of their network, increase network capacity to match user demand. enhance the user experience, and lower the cost of delivering mobile broadband services.

Crosstalk and Copper Wires

Electrical current flowing through any conductor (like the copper wires connecting his phone) will produce a surrounding electromagnetic field. If another conductor is within the surrounding field, an inductively coupled current will flow through the adjacent conductor.

Inductively Coupled Electromagnetic Flux

In the figure above current flowing through the conducting wire will produce an inductively coupled current in the adjacent wire. If the varying signal current represents a voice transmission the conversation can crossover from one line to another and voices can be heard on one line from another line conversation. Usually this is only an annoyance since crosstalk signal levels are typically low when compared with the signal levels of the conversation on the primary line. On the other hand digital data transmissions are extremely sensitive to crosstalk. Crosstalk can cause bit misinterpretation and will typically require a retransmission of the damaged data.

There are two types of crosstalk, near end and far end.

Near End Crosstalk (NEXT)

Near end crosstalk occurs between a transmitted signal and a received signal. Transmitted signals are typically stronger that a signal that is being received and interfere with the received signals.

Near End Crosstalk

Far End Crosstalk (FEXT)

Far end crosstalk occurs between two signals transmitted in the same direction. The adjacent conductors each produce a magnetic field and can interfer with each other.

Far End Crosstalk

The most common way to reduce crosstalk between adjacent wires is to twist the wires together in a way that cancels the crosstalk flux. That's why Unshielded Twisted Pair (UTP) cabling is used for high speed data cabling like Ethernet. In addition shielding, in the form of foil or metallic braid is also used in Shielded Twisted Pair (STP) cable.

Global Mobile Stats - Google Our Mobile Planet

In a project called Our Mobile Planet, Google's been collecting mobile stats from 27 countries. Dan Swinhoe from IDG Connect Global has picked this data apart and written a very nice post titled The App Revolution: How this Varies By Market. Here's a few interesting tidbits from Swinhoe's excellent piece:

• Japan is the most ‘appy', but Germany is amongst the most keen to pay.
• According to 148apps, the Apple store has 719,452 apps available, and to buy them all would set you back a hefty $1,307,715.69.
• Angry Birds Star Wars is currently dominating the App store charts.
• By the end of the year, over 45 billion apps will have been downloaded - around 15 billion of those from Google, but you can expect Android to take the majority share in 2013 due to the sheer number of devices being sold using the search engine's OS. 
• Microsoft's own appstore is yet to make significant inroads in any market but, depending on the success of its Surface tablet this could well change after Christmas.
• Custom-app building continues to grow, today's estimates putting the average cost of development at around $30-40,000. 
• According to a report by Appaccelerator, Apple has become the chosen platform for enterprise app development, with 53.2% of developers picking iOS for corporate app development.
• In all the charts, no matter what system or country, games feature heavily in both free and paid for.
• While things such as social media and certain business software are now fully-apped, other areas are still a while off. For example media outlets are still struggling to cope with apps (web is still a struggle for many), while the largest programs - CAD/CAM and other large engineering/graphics programs simply are too big and complex for apps and mobile devices. At least for now.

Be sure to check out Dan's full post linked here and also take a look at Google's Our Mobile Project - all pretty interesting stuff.

Data Transmission on T1 Carriers - Part 2

In Part 1of this topic I described how a T1 carrier is used to transmit data. Data transmission by nature is "bursty" meaning large amounts of information are typically transmitted and then followed by relatively quiet transmission periods. This can cause transmission problems for T-carrier systems since they rely on timing synchronization. Let's take a look how this potential problem is avoided.

Data Transmission on T-1 Carriers Part 1

Back in December I wrote a post here titled T1 Lines - What They Are. In the post I discuss the Digital Signal (DS) Level System and how combining the equivalent of 24 DS-0 voice channels along with overhead consisting of timing and synchronization bits brings the DS-1 bit rate to 1.644 Mbps - that's a T1. In this post, let's have a look in more detail to get a better idea of how the entire system works. 

The T-1 Carrier uses time division multiplexing and was designed for voice call transmission. When used for data one would think it would be possible to achieve a data bit rate of 64 Kbps over a T-1 carrier. Looking a little closer one sees that data on T-1 carriers is transmitted in the form of only 7 bit words, all eight bits are not used. Why? 

Remember the T carrier system was initially designed for voice. The first signal synchronization used for the T-1 carrier substituted a single in band signaling bit, used for control, for each of the 24 channels in every sixth frame. This means in the sixth and twelfth frames of every T-1 carrier master frame there is a bit used for in-band signaling. This is referred to as bit-robbing. Bit robbing is usually not a problem when transmitting voice. Even though the signal is slightly distorted, the listener on the receiving end cannot perceive the distortion. However this is a major problem when transmitting data as any data received with missing bits will be distorted and received incorrectly. To eliminate the problem caused by bit robbing data on the T-1 carrier is limited to seven bits per frame in all frames. By decreasing the number of bits transmitted the data bit rate is reduced.

For this reason, 56 Kbps Clear Channel Capability is the term used to refer to the T-1 carrier single channel maximum data bit rate.

T-1 Carrier Pulse Cycles

If we look closer at a T-1 Carrier signal we see there are negative and positive pulses combined in the digital pulse train. A sample T-1 signal pulse train is shown in the figure below.

Sample T-1 Pulse Train

It has been found that alternating positive/negative pulse trains (bipolar) produces fewer transmission errors than all positive or all negative pulse trains. These pulses are used to represent binary 1’s and each pulse, when non-zero, is positive half the non-zero cycle (50%) and negative half the non-zero cycle. We can look at an example of a positive (cycle 1) and negative (cycle 4) pulse from the above figure.

Sample T-1 Positive and Negative Going Pulses

In the figure above, T represents the period, or time it takes to complete a single pulse cycle. We can calculate the percent duty cycle using the following equation:

The pulses here are not zero for one half of the pulse period and have a 50% duty cycle. Let’s go back now and look at the original pulse train diagram and look at each cycle:

You can now see that if a pulse is present within a cycle time slot, whether positive or negative, it represents a 1 bit and if no pulse is present, it represents a 0-bit.

In Part 2 of this series I'll cover something called Bipolar with Zero Substitution (B8ZS) for T-1 signal synchronization.

SONET Packet-Oriented Data Framing

In my last legacy PSTN post I discussed how Synchronous Optical Network (SONET) is used to multiplex, transmit and then de-multiplex voice calls. Today, let’s take a look at how SONET  is being used to transmit packet-oriented data (in today’s world - basically Ethernet).

In that last SONET post we said the SONET international equivalent is called Synchronous Digital Hierarchy (SDH). Now, when we talk about data at the SONET/SDH level we’re talking frames (think layer 2 OSI model) and the base unit of framing for SDH is something called a Synchronous Transport Module, level 1 (STM-1) with operates at 155.52 Mbps. 

In the post I also said the base SONET standard bit rate is 51.84 Mbps and is referred to as Optical Carrier  (OC) -1 or Synchronous Transport Level  (STS) -1. Now, because we’re talking 3 times an STS-1 and it is concatenated (combined), the base SONET data framing unit (running at 155.52 Mbps)  is referred to as a STS-3c (Synchronous Transport Signal 3, concatenated) which is also referred to as an OC-3c (Optical Carrier - 3c). 

Now that I have you completely confused (!) lets’s talk a little more about packet frames. A typical packet frame consists of a header, payload (the actual data being sent) and some kind of trailer. I like to use a letter analogy to understand what is going on - someone writes a letter (think of the letter as the payload or data). It gets put on an envelope (think of the envelop as the header and trailer for now). At the sending end the letter gets a destination address, a return address, etc and gets delivered. At the receiving end the letter gets opened, the envelop discarded and the letter itself saved and used.

For an STS-3c framing unit, the payload rate is 149.76 Mbit/s and overhead is 5.76 Mbit/s.

If we look at an individual SONET STS-3c frame - it’s  2,430 octets long. SONET systems transmit nine octets of overhead and then 261 octets of payload in sequence. This transmission is  repeated nine times in 125 micro-seconds until 2,430 octets have been transmitted. 

Timing is critical here (that's why it's called synchronous) for communications across the entire network.

Locking In To An LTE Provider

Amrisa Bhagwandin has an interesting post over at goingLTE.com titled Verizon Reserving Its Phones for Its Own Network?

In the post, Amrisa speculates that Verizon Wireless is designing its phones so they will only run on the Verizon Wireless network. Bhagwandin also speculates AT&T may end up doing the same. Here's some of the technical details:

The Verizon Wireless and AT&T 4G Long Term Evolution (LTE) networks run on different frequency bands:

• AT&T runs in the 704-746 Mega Hertz (MHz) band

• Verizon Wireless runs in the 746-787 MHz band

There is some slight overlap between the two bands but there is not enough overlap for devices to run on each others networks. It's also important to remember the 4G conversion is not going to be like throwing a switch. Tower antennas will be gradually updated from 3G to 4G. This means 4G phones  have both 3G and 4G radios in them - the 4G radio is used when 4G service is available and the 3G radio is used when 4G service is not available. This fallback also causes a problem. In locations where 4G service is not available, Verizon phones will fall back on the Verizon wireless CDMA 3G network and AT&T phones will fall back on the AT&T HSPA/GSM 3G network.

And..... it gets even more complicated - both Verizon Wireless and AT&T both own spectrum through MetroPCS and Bhagwandin thinks we'll see both companies setting up sales through MetroPCS to try and lock customers into their networks . In addition, we may see similar deals being made with Lightsquared and Cricket since both of these companies are developing their own 4G networks.

I'm in wait-and-see mode right now and not going to lock into any new long term wireless contracts until I get a better idea of how it is all going to shake out.

Bridged Taps - More On The Local Loop

A bridged tap is an unterminated wire pair that sits in parallel to the main wire pair. Ideally, the local loop is a continuous wire point-to-point connection. At one time, the local loops were all setup this way but, with the growth of neighborhoods, new unused wire pairs got added. Typically, extra pairs are included though not initially used when cable is run down a street. When a new house is built, or a line is added, a phone company technician taps into one of the unused pairs. The technician typically does not cut the pair, the wires are just “tapped,” leaving the unterminated ends running down the street. This way, if the line is no longer needed, a technician can come out, remove the tap and still use the pair for another customer farther down the street. This leaves a bridged tap with the tap point being where the technician spliced into the wire pair on the street. 

Bridged Tap Example

Bridged taps can create an impairment to the transmission system. A signal on the loop moves down the un-terminated cable and will reflect back to the main pair affecting the main signal. A bridged tap will typically not be noticed at voice transmission frequencies because the wavelength of voice frequencies is always greater than the line length. All that is experienced is a slight increase in attenuation due to added capacitive load which is usually so small it is not detected by the human ear. 

However, when it comes to Digital Subscriber Line (DSL) technologies, bridged taps can cause major data communications problems and frequently require cleaning up by telecom technicians. I'll discuss how DSL technologies work in a future post.

Loading Coils - More On The Local Loop

Early in the development of the telephone system infrastructure designers realized our everyday speech lies in between 125 Hz and 8 KHz with most voice centered between 400 and 600 Hz. With more studies designers realized that humans can recognize and interpret voice if they stayed within this frequency range. Voice frequencies below 200 Hz and above 2 KHz play very little role in voice recognition.

Frequency Range Diagram 

Since the early 1900’s the infrastructure has been tuned to match these frequency requirements using devices called loading coils.

Both George Campbell at AT&T and Michael Pupin at Columbia University were working in 1899 on wire pair mutual capacitance problem. Both realized that, by adding a lump series inductance called a loading coil, resonance could be used to cancel the effects of shunt capacitive reactance and increase signal strength over long local loops. Michael Pupin ended up getting the patent and by late 1899 loading coils were being installed in the field on copper wire pairs longer than 3 miles.

Western Electric 25 Pair Loading Coil Cable Case (circa 1977)
[and.... circa 1977 is the cable case, not me!]

Loading coils are a simple lump series inductance that produce an effect called loading. Loading increases the series inductance of the loop and effectively makes the loop a low pass filter, increasing the impedance of the line which drops signal attenuation. A typical 26 gauge local loop pair is loaded with a 26H88 loading coil. The letter H designates a coil that is added every 6000 feet, 26 represents 26 gauge wire and 88 indicates the inductance of the coil is 88 mH. This loading makes the loop perform as a low pass filter and cuts the frequency off sharply at around 3.4KHz. Loading coils  work great for the low bandwidth requirements of voice but causes problems when you want to transmit data at higher bandwidths over these same wires.

Loaded and Unloaded Loss

 

At voice frequencies, the cutoff frequency (f_C) for a transmission line can be approximated as follows:

     where:    L = Loading Coil Inductance

                       D = Distance in miles btwn loading coils

                       C = capacitance per mile

Example 1

Calculate the cutoff frequency and sketch the frequency response curve for a local 3 mile loop using 26H88 loading coils spaced every 6000 feet. 

Solution:

                                                               L = 88mH 
                                                  D = 6000 feet ≈ 1 mile 
                                                  C = .083 μF/mile                                              

Notice using this formula total loop distance is not used in the calculation – only distance between coils is required.

In addition to H (6000 ft) load coil spacing, there are also B (3000 ft) spacing and D (4500 ft) spacing loading coils. By changing coil spacing along with coil inductance the loop cutoff frequency can be adjusted or tuned to the proper value. Let's look at another example.

Example 2

22mH loading coils are spaced every 3000 ft on a local loop. Calculate the cutoff frequency.

Solution:

                                                                    L = 22mH
                                                      D = 3000 feet ≈ .5 miles
                                                      C = .083 μF/mile

Notice in this example by reducing the distance between coils and decreasing the individual coil inductance values, we can increase the cutoff frequency. There are three commonly used loading coils in the United States and the coil specifications are listed below:

Loading coils have been used over the last 100 years and are an excellent way to tune a local loop to voice frequencies between 300 and 3300 Hz. As carriers move to provide high bandwidth data services such as ADSL on the same local loop being used for voice the low pass filter characteristics of the loaded local loop provide significant bandwidth limitations. We can see frequencies above 4000 Hz on loaded loops are blocked. For this reason loading coils are being removed from the local loop.

Transmission Lines and the Local Loop

I know this post gets a little mathematical. Try and think of the math in simple terms - in the examples below we're dealing with some basic division:

Answer = Numerator / Denominator 

That's numerator (top number) divided by denominator (bottom number) in the equation.

If the numerator is large compared to the denominator then the answer is going to be relatively large (think big number divided by small number gives big number answer and remember...... everything is relative :) ). And vice versa - if the numerator is small compared to the denominator then the answer is going to be small ((think small number divided by big number gives small number answer).

This should help to understand the examples below.

----------------------------------

In my last post I wrote about the local loop - that pair of copper telephone wires most of us still have coming into out homes.These wires have been used for voice in some places for close to 100 years and now, using DSL technologies, to deliver voice and data. AT&T UVerse is even using the local loop to deliver triple play services - voice, video and data. In this post, let's take a little close look transmission lines.

The local telephone loop (also referred to as the subscriber loop) is the dedicated copper wire twisted pair connecting a telephone company Central Office (CO) in a locality to a customer home or business. The loop resistance is critical in the local loop and phone companies have had to “tune” the loop to transmit high-quality voice. Typically, companies have used 19 gauge (1.25 decibels [dB] attenuation per mile) to 26 gauge (3 dB attenuation per mile) copper wire for the local loop. The average customer local loop is about 2 miles and attenuation on this loop is ideally kept below 8 dB.

We can look at a typical transmission line model and use it to represent a subscriber loop:

Transmission Line Model

We can see that the inductance (L), resistances (R for series resistance and S for shunt resistance), and capacitance (C) are distributed throughout the model. We can also show that these values cause signal loss and distortion.  A local loop copper wire pair effectively forms a capacitance since you have two conductors (copper wire) separated by an insulator (wire insulation). Shunt or mutual capacitive reactance is independent of wire gauge and local loop wire pairs designed for voice have a capacitance value of about .083 μF/mile.

In addition to local loop cable, copper cables designed for higher frequencies like those used for T carrier systems are designed to provide a capacitance of .066 μF/mile.

Two Wires Separated by Insulation Forming a Capacitance

Capacitive reactance is basically the resistance of a capacitance and it changes with frequency.   The formula for capacitive reactance is:

The units for capacitive reactance are Ohms (Ω). Looking at the formula you can see as frequency increases the denominator gets larger so the capacitive reactance drops. On long local loops (3 miles and greater) shunt capacitance values increase to the point where significant signal leakage occurs at frequencies greater than 1000 Hz. If you look at the formula, you realize the higher the frequency the greater the leakage loss. Let’s look at some examples:



Example A

A local loop is 1 mile long. Calculate the capacitive reactance for the loop at 2KHz.

Solution:
Using          f = 2 KHz     




Example B

This same local loop is extended to 3 miles. Calculate the new capacitive reactance for the loop at 2KHz

Solution:
Using          f = 2 KHz  




In the example you can see that, by increasing the length of the loop by two miles, shunt capacitance drops by a factor close to 10.

In addition to length, higher frequencies also cause shunt capacitance reactance to increase.


Example C

Let’s increase the frequency in Example B to 3KHz and calculate the capacitive reactance of the local loop.

Solution:

Using          f = 3 KHz    




Example D

Let’s now decrease the frequency to 1KHz and calculate the capacitive reactance of the local loop.

Solution:

Using          f = 1 KHz



Now consider a voice conversation on the Example C local loop. We know the frequency range of the local loop is approximately 300 Hz to 3300 Hz. We know the human voice can produce frequencies of both 3KHz and 1KHz and the average ear can hear these frequencies. At 1 KHz we have a shunt capacitive reactance of 639Ω and  at 3 KHz we have a shunt capacitive reactance of 213Ω. You can see more signal is lost due to capacitive shunting at the higher frequencies than at the lower frequencies. When it comes to voice - the listener will notice these differences – the lower frequencies in a voice conversation will appear louder than the higher frequencies in a conversation.

Over 100 years ago telephone companies figured out they could "load" a transmission line with inductors (loading coils) to reduce the effects of capacitive reactance. I'll discuss loading coils in a future post.