Wireless Network Congestion

There is no one easy fix. Rather, it will take more spectrum, more cell sites, more backhaul, more off-loading of data to Wi-Fi and other local-area technologies, and better behaved applications to help minimize the demand issue. The key word is minimize since there is no “cure” for wireless network overload.

This is a subject I have been talking about for a while now but it is rearing its head again since the latest stats are out: Smartphone data usage is up 100% year over year, the iPhone 4S uses more data than any preceding iPhone, developers are making use of the wireless signaling channels for data updates, and streaming video demand is soaring. The network operators, even those that have rolled out 4G LTE networks, are concerned about the demands on their networks for data services as is evidenced by their moving from unlimited to capped monthly data plans, and more, different types of plans are forthcoming.

Network operators want to provide each customer with equal access to broadband data all of the time but with the demand growing so rapidly, they are hampered by network capacity and the time it takes to build new cell sites closer together, so they are trying other ways including off-loading data services to Wi-Fi and femtocells located in a customer’s home and, of course, initiating pricing models that will slow down the growth rate so the network operators can catch up. I am often questioned about why there are capacity restrictions for wireless networks and I have written a number of articles about this before. However, with the renewed clamor about wireless broadband capacity, I thought it would be worth restating the issues.

Wireless Bandwidth Is Shared Bandwidth

In order to better understand the issue, let’s start at the end-user and work our way into the network. Here are a few basics: The wireless link between a user’s device and the network is, in reality, the last mile, or ½ mile, or the distance from the user device to the nearest cell site. Once this communications link is active, the balance of the voice and data handling is completed within the network as the cell sites are connected to the network with high-speed wired, fiber, or microwave links.

The first bottleneck is this wireless link and here is why. Typical cell site coverage is a basic circle that extends around the cell site for ½, 1, or even up to 3+ miles. In order to provide more capacity per cell site, each site is usually divided into three pie-shaped coverage areas radiating 120 degrees from the site outward. The amount of radio spectrum that is licensed to the network operator and that is deployed within that cell site determines the overall cell site capacity. For this example, the outbound capacity is 5 Mbps for each sector and 3 Mbps of capacity for the communications link from the user device to the cell site. Therefore, the total capacity of that cell sector is 5 Mbps times three sectors or 15 Mbps down and 3 Mbps times three sectors up or 9 Mbps up.

Since a user is assigned only to a single cell sector at a time, the total available capacity is 5 Mbps down and 3 Mbps up. Therefore, if there is only one user within that cell sector, this user will have access to all of the capacity—in this case, 5 Mbps to the device and 3 Mbps up from the device. If there are a number of users within a single cell sector and they are all engaged in email, Internet surfing, and other activities that require intermittent data transmissions over the network, the perception will still be that all of the users have a very good data throughput as the total data capacity will be shared on an as-needed basis.

The issue with network capacity comes into play in one of two ways. The first is when there are too many people within the same cell sector, and the second is if some of those within the cell sector are streaming video either up or down. Since the total bandwidth is shared, if one user is watching a video that is being streamed at say, 2 Mbps, then the other users within the cell sector have effectively lost that 2 Mbps of capacity in the downward direction.

If three or four of the users within a cell sector decide to watch a streaming video, it is easy to see that the cell sector could run out of capacity. In this case, those who started watching their video first would retain some bandwidth while others would see their video slowing down, breaking up, or not being able to be streamed at all.

Newer technologies such as 4G LTE employ a number of methods in an attempt to keep up with cell sector demand. Quality of Service (QoS) is available and in some cases network operators can invoke priority service restrictions to favor some users over others. But the outcome at some point is the same. If the demand for uplink or downlink data within a single cell sector is exceeded, some or all of the users within that sector will experience much slower delivery of data or they may be denied service altogether.

Moving to another cell sector that is less congested will restore the data speed and capacity.

The over-demand for data within a single cell sector is the main factor that causes users to experience slower data rates or not be able to connect to their data at all. There are other factors that can cause the same problems, and some of them will affect not only a single cell sector but perhaps a group of sectors. Some network operators have added the capability to expand or shrink the size of a cell sector so that traffic in the overloaded sector can be routed into a less busy sector. This can be accomplished in many ways including the use of antennas that can be electronically tilted down or up. In this type of system, a busy cell sector might be made to appear smaller and the surrounding sectors made to appear larger, which results in a redistribution of traffic among the three sectors. In the newer network technologies, much more flexibility for real-time sector management has been built in and can be employed when needed.

Signaling Channel

In order for a wireless device to be able to operate on a network, the device must be registered on the network and the network must know in which cell sector the device is located at any one time. This form of communications between a device and the network is usually over what is called the signaling channel. This channel’s purpose is to keep track of the devices within a cell sector and to be able to route calls and traffic to and from the device. This channel is also used when a device is moving from one cell sector to another to help accomplish the handoff between the sectors. As a user’s signal level in one cell sector diminishes it will become “louder” in another sector. At exactly the right point, the network will move the device from being associated with the first cell sector over to the second sector. This happens so quickly that you don’t know it is happening and you don’t lose any voice or other information. In normal operation, as you move from sector to sector the user device will talk to the network on an ongoing basis and the handoffs will occur even if the device is on but not being used.

Depending on the wireless technology deployed, the signaling channel can also serve other purposes. For example, in some systems the signaling channel is where the text messages or SMS traffic is routed, and some network developers have found that they can also use the signaling channel to update the data in near-real time on the device. Both of these, as well as other uses of the signaling channel, can reduce the capacity. Think of the cell sector’s broadband capability as a fire hose spewing out lots of water over a given area while the control of that data and the devices is being handled by a garden hose. The problem is that if the garden hose becomes overloaded, the traffic to and from the network from devices may not actually reach the network. If the network cannot communicate with a device, essentially, the network does not know that a device is trying to connect.

This is exactly what happened in the earthquake on the east coast. All of the networks were up and running but too many people in on one area were trying to communicate with the same cell sector or site, overloading the signaling channels. Calls were blocked, text was blocked, and data services were blocked. Those that did access the network might have been disconnected and then could not reconnect. This was simply a case of network overload.

Leaving the Cell Site

Once your voice, text, or data call reaches the cell sector it is then carried via wire, fiber, or microwave into the network where it is then routed to where it is supposed to go—to another phone, the Public Switched Telephone Network (PSTN), the Internet, or wherever. Typically, cell sites are designed to be operating at less than capacity. If the cell site is operating close to capacity, the amount of capacity available to transport information from the cell site to the network might not be sufficient to handle all of the data. Another way to look at it is if the site (all three sectors) is capable of handling 15 Mbps of traffic (3-5 Mbps sectors), the backhaul might be designed to handle only 10 Mbps rather than the full 15 Mbps since 99% of the time all three sectors would not be loaded to capacity.

In this case, even though there was capacity available on the wireless portion of the cell site, the backhaul would not be able to transport all of the data to and from the network and the result would be the same as if all of the wireless capacity was being used. If this were the case with a number of sites in the same congested area, then those who could not access the network would be spread out over multiple cell sites. With the advent of 4G LTE, this situation is far less likely to happen since the backhaul for many 2G and 3G sites was a number of wired phone circuits while 4G networks are using fiber and microwave with much more bandwidth. Still, if the network operator does not plan for more than average data usage at some sites and skimps on the capacity of the backhaul, the result could appear to be network congestion.

Conclusions

Cellular networks were designed to be able to cell split. You might start with cell sites five miles apart from each other and as traffic increases you build a new site between the first to and then divide the area again and again. This is a great way to add capacity without having more spectrum available and it worked well in the early days of cellular systems. Today, with the public wanting coverage but not a cell site located near them, the time it takes to identify a new site and build it out has gone from a few months in the 1990s to 18-24 months or longer today. Further, the cost of a full-blown cell site is in the multiple $100K range or more depending on the location and the type of construction.

Network operators are using Wi-Fi, picocells, microcells, and femtocells to add to their capacity. The Federal Government is working diligently to find more spectrum so operators can have more, but this too is a long-range goal. In the meantime, as the demand for data continues to double year over year and those who are writing applications for these devices are unaware of the issues assuming that bandwidth will always be there, the problem of wireless network congestion will continue and in fact will become more prevalent. This is one reason network operators have gone to usage-based contracts.

There is no one easy fix. Rather, it will take more spectrum, more cell sites, more backhaul, more off-loading of data to Wi-Fi and other local-area technologies, and better behaved applications to help minimize the demand issue. The key word is minimize since there is no “cure” for wireless network overload. Regardless of how much more capacity we create, if the demand continues to increase, the issue of network congestion will continue to be with us.

Andrew M. Seybold

 

4 Comments on “Wireless Network Congestion”

  1. InfoStack says:

    This is a good analysis, but it is supply centric. There is a demand element and addressed properly via software (applications, cueing) and marketing (pricing and billing) constructs can be modified to reduce congestion significantly.

  2. InfoStack says:

    Have you seen an analysis of how much data goes through wifi vs macro-cellular already today? In its entirety, I wouldn’t be surprised if 85-90%+ of all wireless data goes through wifi and less than 10% through macro-cellular. This is simply the market recognizing that WiFi is government mandated (nearly) infinite spectrum reuse.

  3. […] Mobile industry analysts now systematically include in network congestion coverage – Chetan Sharma, Andrew Seybold […]

  4. […] transmission from reaching unintended devices. The only thing a network operator can do to relieve congestion is to divide the network up in geographic segments and limit the power at each tower from […]

You must be logged in to comment or reply.