The traditional way to establish network connectivity simply involves connecting wires to the Internet or the enterprise network. During the past few years, though, the use of Wi-Fi IEEE 802.11 technology has opened the door to wireless-based connectivity. Such connectivity is commonly deployed in hotspots, with access points (APs) located near populated areas like conference rooms or meeting rooms.

Recent market developments have introduced two major network requirements:

• Full wireless coverage: A large percentage of end-user equipment isn’t intended for wire connectivity (e.g., tablets and smart phones).
• High wireless capacity: Storage became more centralistic, on high-volume servers, instead of local storage. More users and multimedia applications intensify demand for Internet access.

These two requirements introduce new challenges for wireless indoor networks using high-capacity IEEE 802.11n multiple-input multiple-output (MIMO) technology with full enterprise coverage. However, recently introduced concepts help overcome these unique indoor Wi-Fi deployment capacity issues.

For instance, Wi-Fi over Distributed Antenna System (DAS) provides a solution for all IEEE 802.11 communication, including IEEE 802.11n MIMO, throughout indoor enterprise environments. It delivers centrally managed Wi-Fi APs that also solve the “hidden node” problem, resulting in higher overall capacity capabilities.

High-Capacity Coverage Challenges

Typical indoor Wi-Fi deployments include many APs located throughout the building. Increasing capacity demand requires a higher signal-to-noise ratio (SNR) and APs to move closer to one another to create smaller cells. For example, in a Wi-Fi deployment with full high-capacity coverage, SNR is high enough to provide maximal physical-layer (PHY) rate in every location throughout the enterprise (Fig. 1). The circles define the deployed APs, while their color defines the three non-overlapping industrial, scientific, medical (ISM) 2.4-GHz RF channels.

Measuring power received in different locations revealed that three or four APs were received in most locations. Thus, it’s practically impossible to create non-overlapping hotspots that cover an entire area in most indoor environments.

An area in which high capacity is available and a larger area that maintains lower capacity surround each of these APs (Fig. 2). The lower-capacity area is defined herein as an interference area, which is caused by interference from the adjacent AP.

Increasing capacity demand and APs in closer proximity creates greater interference between cells. Due to the Wi-Fi carrier sense mechanism, an AP or station (co-located in the same RF channel as another AP or station and able to receive its packets) will deny transmission as long as the channel is busy.

Also, in the case of many APs and stations, there’s a higher probability of a collision between data packets or between data packets and acknowledgement (ACK) packets. When collisions take place, the interrupted transmitting unit will select a backoff time and retransmit the same packet. As more collisions occur, system efficiency degrades further, decreasing total capacity.

The simulation plot demonstrates the total capacity of IEEE 802.11n with MIMO when adding more transmitting APs on the same wireless media (Fig. 3). In this scenario, capacity is expected to grow, since the IEEE 802.11 protocol is carrier sense multiple access (CSMA) and because units don’t transmit simultaneously. However, total capacity drops due to collisions and interference between links.

An experiment conducted by Alvarion revealed such a capacity decrease. Eight stations were deployed in an indoor environment, transmitting 3 Mbits/s of TCP traffic to each one (see the table). When using one AP in high SNR conditions, total system capacity equaled 14 Mbits/s and each piece of customer premises equipment (CPE) received about 1.75-Mbit/s throughput. With four APs, under the same high SNR conditions, total system capacity dropped to about 9 Mbits/s. In fact, some stations were starved (i.e., negligible throughput), while others even disconnected.

Wi-Fi Over DAS

DAS can reduce the number of APs without decreasing station SNR. Antennas are deployed throughout the enterprise and connected to the AP physical antenna ports. The connection between the AP antenna ports and the distributed antennas may be passive (cables, splitters, etc.) or active (amplifiers, frequency conversion, etc.).

By using active DAS, just one coax cable can distribute the signal among the different antennas. Each antenna connects to the DAS network via a remote unit, which filters the relevant signal from the coax, converts it back to the original signal, and transmits it using a power amplifier. As a result, the total antenna distribution doesn’t suffer any power loss. Moreover, several MIMO channels can be sent over a single coax cable by converting each channel to a different frequency and converting the signal back to the same frequency at the remote unit.

Hidden Nodes

Although Wi-Fi DAS is a good practical method for improving station SNR and total system capacity, connecting the AP with remote units could create challenges such as hidden nodes, which occur when Wi-Fi stations can’t sense one another. Thus, when performing uplink traffic, packets from hidden stations keep colliding with one another, resulting in poor throughput.

Such a scenario occurs frequently in simple Wi-Fi DAS deployments. Stations are connected to the AP through different remote units, which are likely to be located far apart and not be able to sense one another. Alvarion developed a method to eliminate the hidden-node problem.

Performance Evaluation

Wi-Fi DAS performance tests conducted by Alvarion included two remote units, each one connected to one Wi-Fi client by a single cable connection. The tests were performed using IEEE 802.11g clients and APs in high SNR conditions, allowing for a 54-Mbit/s PHY rate.

In both trials, client 1 transmitted TCP uplink traffic at maximum capacity to a server connected to the AP. At the beginning of the test, client 2 was set to start sending a TCP uplink to the same server. Momentary throughputs of client 1, client 2, and the total throughput were recorded each second. Each recording included the average throughput of one second.

Results showed that when both clients were connected by a standard Wi-Fi DAS, client 2 was not able to transmit traffic at all for six minutes (starvation) (Fig. 4). After that, the average total throughput (denoted in black) was about 13 Mbits/s, with high fluctuations in the throughput of both clients (even including drops to zero throughput from time to time).

A separate test, using identical parameters, was conducted with Alvarion’s solution (Fig. 5). Adding the second transmission caused the first client’s throughput to stabilize within 15 seconds, while minimum throughput within the transition period was more than 80% of the stable value. The final total throughput (denoted in black) was stable with very low fluctuations, maintaining an average value of 18 Mbits/s.

Overall, the new DAS deployment concept helps bring high-capacity Wi-Fi communication to indoor enterprise environments. Tests show that it increases the wireless efficiency for all IEEE 802.11 communication, as well as boosts capacity, solving two major classic problems in Wi-Fi DAS networks.

All claims made in this article have been tested and proved using standard enterprise Wi-Fi equipment. Test results with two clients showed over 40% improvement in throughput and much higher robustness. On top of that, greater improvements are possible when there’s an increase in the number of clients and remote units (which enhances the likelihood of more severe collision problems).