Upload
haxuyen
View
217
Download
1
Embed Size (px)
Citation preview
ARUBA 802.11AC LAB TESTING PROJECT REPORT
May 3, 2014
Product Category: Wireless Access Point
Vendor Tested: Aruba:
Products Tested: Aruba Networks AP-‐225
ABSTRACT This is a report to understand the parameters to implement a wireless network in an enterprise. A performed set of experiments is used to explain few important parameters and an attempt is made to understand the feasibility of a network.
Aniruddha Nandam Bhavik N. Patel Manish Sadhwani IST 648 Enterprise Wireless Network
EXECUTIVE SUMMARY:
Most of the enterprises today are implementing wireless networks for their operation or are in the phase of implementing such. While wireless networks bring about mobility in such deployment situations, they are also not as efficient in terms of throughput as compared to wired networks. However, considering the present technology it might not be huge of a problem depending on the method in which it is being implemented. This is when the complexity of implementing wireless networks enters. While wireless networks seem to be a transparent ease of use implementation in terms of users, it is much harder for network administrators to implement and monitor such networks. It is further difficult to understand their ideal configuration or best-‐suited requirement configuration. This paper targets this issue in not all but few of the most important areas of implementation.
While implementing wireless in an enterprise network it is important to understand the feasibility of the network and to understand the specifications, which might be the best to implement the network considering the requirement of the enterprise. This includes the channel width to be implemented, primary channel to implement the network, other specifications to be decided considering the vendor whose products are to be implemented. More important is to know the range of the AP to understand the coverage of the network, effect of the type and number of clients, which are accessing the network. Finally it is important to understand the performance of the network and monitor it in real time. In this project we have tried to understand the effect of these parameters using various tests and documented these results in the form of graphs for easy visual realization.
While this report consists of major issues to be considered while implementing wireless networks, only the configuration related to the test scenario is modified while maintaining other specifications as default. This might not be perfect implementation of an enterprise network and would vary from situation to situation depending on the requirements of the network.
Test Objectives:
In this report we have included the results on tests performed to understand the effect of few parameters related to implementation of enterprise wireless network.
In the experiments that we performed we considered the following specifics:
• Rate Verses Range: It is important to understand the range of the AP such that it would provide the throughput requirements of the network users. While this explains the range of the network, a quantified set of readings can help understand the placement of the AP. The set of experiments we performed also tried to understand the effect of intrusion of different type of devices and trying to explain the effect of throughput received on one client while increasing the number and type of client.
• Multi-‐client testing: This set of testing was performed to understand the situation in which the increasing number of clients would behave in an environment. This would be an ideal situation in which one of the clients is associated with and AP, However as newer clients are added to the network the throughput would be affected due to sharing of radio resources and effect of channel share time on these devices leading to change in throughput received by each client.
• Aruba Adaptive Radio Management: Aruba provides a smart radio management and intelligence, which is used to monitor and manage radio transmission parameters. An attempt to understand this feature is done in this report.
While the above set of core parameters are tried to be understood in the project, we performed a basic comparative set of tests to decide the specifics of configuring the AP to perform the above tests. This included the channel width and upstream/downstream behavior.
TESTING TOOLS:
Hardware Used:
Aruba AP 225:
• Has maximum data rate of 1.3 Gbps in the 5 GHz band & 600 Mbps in the 2.4 GHz band • Eliminates sticky clients with the use of Client Match Technology • Can function on 802.11af i.e. POE.
Aruba 3600 controller:
• Can support up to 128 APs or Mobility access switches • Manages authentication, encryption and VPN connections and Layer 3 networking • Contains Aruba Policy Enforcement Firewall (PEF), Aruba Adaptive Radio Management
(ARM), and Aruba RF Protect™ Spectrum Analysis • Contains Airwave that provides real-‐time monitoring, reporting and troubleshooting.
Meraki MS22P Switch:
• Contains 24 Gigabit Ethernet ports. • Supports POE+ i.e. 80.11at • Supports voice and video QOS • Real-‐time troubleshooting
Cisco 3760G Switch:
• Password protected for secure access • Supports POE and POE+ • Supports Auto-‐QOS for easy QOS implementation • Supports Cisco Energy wise for energy management of POE
Dell PowerEdge R200:
• Provides flexibility in performance depending on the load • The server provides scalability and can be easily upgraded • Contains management tools for troubleshooting
Software Used:
iPerf/jPerf software was used for obtaining the throughput for single client range vs. rate tests as well as for multi client tests. iPerf is a tool to measure network performance for measuring bandwidth provided by the network. iPerf is a tool to measure maximum TCP bandwidth, allowing the tuning of various parameters and UDP characteristics.
The screenshot above shows the jPerf client setup for executing the test scenarios. As seen in the screenshot, on the client side, the IP address of server was entered keeping the port no 5001 and number of parallel streams 5 or 6 as appropriate for that specific test scenario. Each test run was conducted for 30 seconds and the output was set to be displayed in Mbps.
For measuring the signal levels at different locations on the floor and to identify the number of channels for free transmission, we used spectrum analysis tools called Chanalyzer Pro and InSSIDer installed on a Dell laptop. From the spectrum analysis, RSSI was observed at various locations and finally seven locations were selected for testing. Also, it was decided to run all the tests with the AP transmitting on the channel 149 + 151.
TESTING ENVIRONMENT AND CONFIGURATION:
Network Diagram:
The figure above shows the network setup for testing that comprised of an Aruba Wireless LAN Controller, Layer 3 Cisco switch, DHCP server, Aruba Access point and single/multiple clients. Aruba AP 225 can be exploited completely on its full power mode only via POE+ cable. Hence, another Meraki switch was also introduced in the network to provide POE+ support to the AP. We used an HP ENVY m7-‐j010dx Notebook with 10/100/1000 Gigabit Ethernet LAN (RJ-‐45 connector) as the server for testing with jPerf. A jPerf/iPerf server was created on this machine connected to the Aruba Wireless 3600 series controller using a CAT 6e cable via Meraki MS22P Switch provided with DHCP services from Dell PowerEdge R200 acting as DHCP server. The wireless clients were connected to the Aruba AP 225, also connected to the Aruba Wireless
3600 series controller using similar arrangement as HP laptop obtaining IP address from DHCP server thereby getting connected to the controller.
Basic Testing for Test Configurations:
In order to implement a Wireless network it is important to decide the parameters of implementing the wireless network. This includes the width of channel on which the wireless network is being broadcasted. This can be 20 MHz, 40 MHz or 80 MHz. As also we need to understand the traffic flow requirements of the network, which can be upstream/downstream. In a real life situation the amount of downstream traffic is generally higher than the upstream traffic and it would be important to know the throughput, which can be delivered in either case or a case in which there is upstream along with downstream traffic. Hence, a small set of tests was performed to understand the effect of these parameters on the network to be setup.
20/40/80 MHz Comparison:
In order understand the effect of channel width we configured the access point to transmit in the default specifications of channel 149 for a 20 MHz channel with the other options set to default. This included selection of options for High throughput enable (radio) and Very high throughput enable (radio) mode. Then a throughput test was performed to observe the performance of network in downstream mode. We received the following set of readings for the same.
Channel width 20 MHz 40 MHz 80 MHz Reading 1 368 219 427 Reading 2 366 236 311 Reading 3 274 231 325 Average 335 230 355
The above set of readings were received at a distance of 20 feet, where we received a signal strength of -‐62 dBm and the transmit rate received was 527 Mbps for all of the above cases.
As you can see from the above set of readings we can see that the effect of channel width is not as expected from theory. This can be reasoned to the default selection of Aruba Radio Management (ARM) and selection of Very high throughput enable (radio) mode. This might be responsible to override the channel width setting for providing throughput according to the nature of the client. The Aruba Web interface displays dashboard, which displays the clients, connected to the network and the capability of it. As an example the MacBook pro client to the
network was recognized as 802.11ac device of 80 MHz channel capacity. Hence, we observe a similar set of readings for either of the selection of channel width of 20/40/80 MHz.
Upstream/Downstream Comparison:
It is also notable to observe the type of traffic and corresponding effect on the performance of the network. In order to perform this we connected a HP Laptop via Ethernet to serve as a JPerf/iPerf client at first to transmit data to server, which gets connected to the network via Wireless connectivity, this simulates a typical downstream flow of traffic on the network. This arrangement was then modified a bit to observe upstream traffic. This was done by keeping the device connectivity as the same, However changing the JPerf setting on HP Laptop to serve as server and setting MacBook Pro 802.11ac client to client to transmit data from wireless medium to the wired connected server. Finally, the parameter on the client (HP Laptop-‐Wired) is set to –d, which corresponds to demonstrate both way traffic and displays both upstream and downstream traffic behavior. This can be represented using the table:
Only Upstream Only Downstream Both way traffic
Upstream Downstream 219 313 126 228 236 299 153 228 231 348 154 209
The above set of readings was received at same distance as above for signal strength of -‐63 dBm and a transmit rate received was 527 Mbps. The following graph illustrates a graphical inference that we understand as explained below.
Graph 1: Throughput Comparison of Upstream and Downstream Traffic
As you can see from the above readings we get a higher downstream rate as expected also when we have dedicated server-‐client scenario and when we have a both way traffic flow between the server and the client.
AP configuration:
Based on the readings obtained in the above-‐performed tests, we configured the access point as described below.
• Tests performed in the 2.4 GHz band used 20 MHz channels, while tests performed in the 5 GHz band used 40 MHz channels.
• AP channel selection was static set to channel number 149 + 151 for 5 GHz and channel number 6 for 2.4 GHz.
• For 5 GHz, ‘High Throughput’ and ‘Very High Throughput’ radios were enabled.
Note: All the readings were taken in Mbps. All the values in the graphs are represented in
Mbps.
0
50
100
150
200
250
300
350
400
Upstream Downtream
Only Upstream Only Downstream Both way traffic
Upstream vs. Downstream
Reading 1 Reading 2 Reading 3
Floor Plan:
Based on the RSSI values obtained during a site survey of the second floor of Hinds Hall, Syracuse University, six test locations were finalized as shown in the figure above. One more test location was added other than those described in the figure above. This was just below the AP to minimize the interference effect. The description of all the locations is as follows:
1. Just Below AP (0 feet) – to observe throughput with minimum interference. 2. Hallway door – considered to be in direct LOS of the AP. RSSI value: -‐36 dBm 3. Hallway (near Room 207) – Another location in direct LOS but at a farther distance.
RSSI value: -‐45 dBm 4. Near Room 210 – Not in LOS, but with minimum interference outside LOS.
RSSI value: -‐51 dBm 5. Conference Room 216 – Going farther away from access point introducing more
interference. RSSI value: -‐62 dBm
6. Near Room 224 – Interference more intense due to intervening walls. RSSI value: -‐73 dBm
7. Near Room 232 – basically to the end of hallway providing maximum interference on the same floor. RSSI value: -‐80 dBm
RATE VS. RANGE TESTS:
Test 1 – Single Client, Single AP Throughput Test: First we started with basic testing by accessing Aruba AP from just one client at a time. The purpose of these single client tests was not any way related to frame a real world scenario for testing the AP. Instead the purpose of this testing was to obtain the actual throughput value of the Aruba AP to form a basic ground for our further testing. As we all know, there are some differences between the data rates of an AP and its throughput. In these tests we had just one client present in AP’s network and trying to capitalize optimally by transmitting all the parallel streams over the complete channel available for communication. Rate vs. range tests would give us a better picture regarding the throughput capabilities of the AP. We selected all 802.11ac clients for these test scenarios. Following were the clients used for single client throughput test:
-‐ New MacBook Pro with Retina display that is 802.11ac compatible and 3x3 MIMO -‐ Same MacBook Pro with Asus Wi-‐Fi Adapter that is 2x2 MIMO -‐ Samsung Note 3 with 802.11ac support and 1x1 MIMO
We chose MacBook for these tests because they support 802.11ac standard and provide slightly higher throughput when compared to other Dell/Intel laptops. Asus wireless adapter is a USB version 3.0 adapter. Hence it would be the best option to choose a device that has USB version 3.0 for testing the Asus adapter, as it would exploit it completely to provide appropriate readings of throughput for a single client test. New MacBook Pro satisfied that requirement and was apt for such testing.
We conducted these tests at all the testing locations described in the floor plan. The RSSI levels at each location were measured each time a test was executed and accordingly decision was made on client’s jPerf regarding the number of parallel streams to be transmitted while running the test. The number of parallel streams was either 5 or 6. Set of three readings was carried out on each client at each location and average of these reading was calculated. This avoided any sudden changes in the environment that may result into abrupt drop or rise in the throughput. The readings were captured till the time they were stable and consistent.
Results:
Graph 2: Throughput results for Rate vs. Range test on New MacBook Pro Client (3x3)
Graph 3: Throughput results for Rate vs. Range test on MacBook Pro with Asus Adapter (2x2)
0
50
100
150
200
250
300
350
400
450
Just below AP Hallway door Hallway (Room 207)
Near Room 210 Conference room 216
Near Room 224 Near Room 232
New MBP (3x3)
0
50
100
150
200
250
Just below AP Hallway door Hallway (Room 207)
Near Room 210 Conference room 216
Near Room 224 Near Room 232
MBP with ASUS adapter (2x2)
Graph 4: Throughput results for Rate vs. Range test on Samsung Note 3 (1x1)
As expected for all the three clients, the maximum throughput was obtained when the clients were in direct Line of Sight (LOS) of the AP. The throughput obtained for New MBP (3x3 client) is highest of all three clients just below the AP with a value of 408 Mbps, whereas maximum throughput obtained for MBP with Asus adapter and Note 3 is 210 Mbps and 185 Mbps respectively.
But, as the distance between the client and the AP increased, the throughput decreased. This drop in the throughput was not only due to the increase in the distance between client and AP but also due to the interference caused by the intervening walls. It can be seen from the graphs of first two clients (MBP and MBP with Asus adapter), that there is sudden drop in the throughput when the client moves near the conference room 216. This drop in the throughput is not so abrupt when the client was Samsung Note. Although Samsung Note (1x1 client) gives a lower throughput value even at LOS locations, there is not much variation in throughput with the variation in distance from the AP. It’s a gradual decrease in throughput when compared to other two clients (3x3 and 2x2). This can be clearly observed in the following graph that compares the throughput values of all the three clients obtained at different locations.
0 20 40 60 80 100 120 140 160 180 200
Just below AP Hallway door Hallway (Room 207)
Near Room 210 Conference room 216
Near Room 224 Near Room 232
Note3 (1x1)
Graph 5: Comparison of Throughput results for Rate vs. Range test on three 802.11ac clients
Since the RSSI value was very low at the farthest location (near Room 232), the throughput values obtained at this location were very less (almost negligible for 3x3 and 1x1 clients). But still the readings could be taken and displayed on the graph as the clients could find the AP successfully and the signal was not dead.
0 50 100 150 200 250 300 350 400 450
Just below AP Hallway door Hallway (Room 207)
Near Room 210 Conference room 216
Near Room 224 Near Room 232
Throughput Comparison of 3 clients
New MBP (3x3) MBP with ASUS adapter (2x2) Note3 (1x1)
Test 2 – Dual Client, Single AP Throughput Test:
In this test scenario, we extended our Rate vs. Range test from a single client to dual client. Here, two clients were simultaneously accessing the AP and jPerf test was run on both the clients together. The clients used for this test case were:
-‐ New MacBook Pro with Retina display that is 802.11ac compatible and 3x3 MIMO -‐ MacBook Pro with Asus Wi-‐Fi Adapter that is 2x2 MIMO
For Asus adapter, MacBook Pro was used because it has USB version 3.0 support that has the optimum compatibility with Asus adapter. Both the clients were given the same server address on jPerf accessing the same AP. Test run were executed on both the clients simultaneously for the same period of time. Similar to single client testing, set of multiple readings was taken on both the clients to obtain a consistent result and average of all these reading was calculated and plotted on a graph. The readings were taken at all the locations mentioned in the floor plan as done in single client testing except the last location – near Room 232. The signal level at this location dropped to a very low level and the clients were unable to reach the AP.
Graph 6: Throughput results for Rate vs. Range test – Dual Client accessing simultaneously
From the graph, it can be seen that the throughput values have declined compare to the values obtained for single client testing using the same clients. This is because of channel sharing between both the clients accessing the same AP at the same time. As expected, the throughput obtained for a 3x3 MIMO client is higher than that obtained for 2x2 MIMO client at each location. The difference between the throughput values of both the clients increases with the increase in the distance from the AP.
0 20 40 60 80 100 120 140 160 180
Just below AP Hallway door Hallway (Room 207)
Near Room 210 Conference room 216
Near Room 224
Dual Client Throughput Test
MBP 3x3 MBP with ASUS 2x2
Test 3 – Triple Client, Single AP Throughput Test:
Similar to the tests performed previously, we further extended our parameters involved in a rate vs. range test by further increasing the number of clients to three. In this case, three 802.11ac clients will access the AP simultaneously and throughput values obtained for all the three clients will be observed. For this we used the same three clients used for single client testing, the only difference being that this time all the clients were running the tests on their jPerf simultaneously for the same period of time. List of clients used:
-‐ New MacBook Pro with Retina display that is 802.11ac compatible and 3x3 MIMO -‐ Same MacBook Pro with Asus Wi-‐Fi Adapter that is 2x2 MIMO -‐ Samsung Note 3 with 802.11ac support and 1x1 MIMO
All the clients were assigned with the same server address on their jPerf and started the test together. Similar to single client testing, set of multiple readings was taken on all the three clients to obtain a consistent result and average of all these reading was calculated and plotted on a graph. The readings were taken at all the locations mentioned in the floor plan as done in single client testing except the last two locations – near Room 232 and near Room 224. The signal levels at these locations dropped to a very low level and the clients were unable to reach the AP.
Graph 7: Throughput results for Rate vs. Range test – Triple Client accessing simultaneously
As observed in the graph above, throughput values have dropped considerably at the last location, near Conference Room 216. Also, when all the three clients were accessing the AP simultaneously, the throughput obtained for 3x3 MIMO client is substantially higher than that obtained for the remaining two clients.
0 20 40 60 80 100 120 140 160 180
Just below AP Hallway door Hallway (Room 207) Near Room 210 Conference room 216
Triple Client Throughput Test
New MBP (3x3) MBP with ASUS adapter (2x2) Note3 (1x1)
MULTI-‐CLIENT TESTING:
In this set of tests, the aim was to observe the behavior of the network when the number of clients or the load was being increased. In order to do this a total of 16 clients were aimed to connect to the same network using the AP utilized in the first suit of tests. In order to simulate a practical situation in an enterprise we considered the following variation of devices to be used as clients. This included:
• 802.11ac Clients -‐ Apple MacBook pro 2013 (802.11a/g/n/ac 3x3 MIMO) -‐ Dell Laptop with Asus 802.11ac adapter (2x2 MIMO) -‐ Samsung Note 3(1x1)
• 802.11n Clients-‐ Mac -‐ Apple MacBook Pro (802.11a/g/n)
• 802.11n Clients Windows -‐ Dell Latitude E6400 (802.11a/g/n 3x3 MIMO)
• 802.11g Clients Windows -‐ Dell Latitude M2400 (802.11g)
These devices were all placed at a same distance from the access point mounted as before. These clients were connected to the AP, broadcasting SSID in 5GHz spectrum in the 40 MHz channel of 149+152. All of these clients were placed at the same distance of 20 feet and were receiving signal strength of -‐50 dBm. However not all of these clients were associated with the AP for the complete duration of the test and were associated for the complete duration of time after initiating its connection to the AP. The clients were placed on a table such that the screens were kept open for a suitable viewing angle (approximately 70º). They were placed such that the AP was facing the rear side of the screen; in other words, the client user was facing the AP directly.
In order to perform this test, we utilized a HP ENVY m7-‐j010dx Notebook with 10/100/1000 Gigabit Ethernet LAN (RJ-‐45 connector) as server. A JPerf /iPerf server using TCP was created on this machine connected to the Aruba AP 225 via wired network. All the clients were then getting connected to the network using this AP and were running JPerf/ iPerf in client mode also using TCP to connect to the HP JPerf server in order to test throughput. All of these clients are scheduled for measuring throughput for the complete duration of test once they are getting connected to the network, which happens one after another in the same order of the list as specified above. Rest of the settings on the JPerf utility is set as defaults for implementing the test. Since, each of them utilizes one stream of data to test throughput via JPerf utility and this gets represented on the server via a channel number associated to it when it gets connected. The HP Laptop server is responsible for measurement of throughput from all of these clients and keeps a track of the Throughput for each client.
This test was performed to simulate the behavior in a real world environment where devices would get connected to a network in short intervals of time. The scenario we used for this deployment was to initially start one 802.11ac device (which was an Apple MacBook pro-‐802.11ac 3x3 device on our case). Associating a Dell Laptop with Asus 802.11ac adapter, which was a 2x2 MIMO after, and interval of 2 min and connecting it as a JPerf client device followed this. Similarly after an interval of 2 min we connected another Laptop with Asus 802.11ac adapter with 2x2 MIMO and Note 3 device. In a similar fashion other legacy devices were connected to the network and tested for throughput by association in a regular interval of around 2 min. All of these devices were initially not associated with the AP and were associated with the AP, followed by starting the JPerf client on it until the end of the test. The results we obtained can be summarized in the following graph:
Graph 8: Throughput results Multiple Clients accessing the AP simultaneously
We can see that initially when only the MacBook pro, which is a 3x3 device, was connected we received a throughput of 330 Mbps. However, if we continue to increase the number of clients the throughput received reduces further and further. While connecting each new device at a regular interval of 2 min we connected 16 devices till the reading 18 point, after which the environment was left to stabilize for about 15 min and due to which we then see again a rise in the throughput values at test point 19. While this set of readings explain a lot about the
0
50
100
150
200
250
300
350
400
450
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19
Effect of MulSple Clients over load
MacBook pro Azuz Dongle Azuz Dongle Note 3
Mac (802.11n) Avg DELL (802.11n) Avg DELL (802.11g) Avg
behavior about the Wireless transmission on increased number of clients it was also worth noting that one of the clients Asus Dongle 2x2 which was the second device to enter the test environment which lost connectivity within a short interval of time and one of the possible explanation for such could be high allocation of transmission time slot for one of the client not providing sufficient resources for other and thereby causing client failure. Which is also one of the major factor while deploying many clients, wherein time slot management becomes increasingly difficult leading to drop of connection for clients trying to get into the network. If we plot the readings of throughput obtained at the initiation and at the end and average of all the readings that we received during the test performed, we can see that:
Clients Throughput (Mbps)
At Start At End Average
MacBook Pro (802.11ac 3x3) 331 96.3 41.91
Asus adapter (802.11ac 2x2) 17 3.73 Samsung Note 3 (802.11ac) 181 25.7
Apple MacBook Pro (802.11a/g/n)
24 10.4
12.6 70.7 14.6 98.3 2.98 53.5 12.8
Dell Latitude E6400 (802.11a/g/n 3x3)
2.62 2.36
2.8675 3.67 3.34 3.41 3.08 3.34 3.28
Dell Latitude M2400 (802.11g)
2.16 5.66
5.643333 8.77 6.03 7.34 8.44 8.06 5.24
From the above set of readings we can observe that the throughput for each client individually decreases while progressing in the test and that increasing the number of clients has substantial drop in the received throughput across the network, this is due to channel sharing which happens across all the devices connected in the network. The following graph represents the aggregate throughput that we receive from all the clients that are connected to the network:
Graph 9: Aggregate of Throughput results for Multiple Clients Testing
As we can see that the throughput we received while fewer clients are connected to the network is higher while as the number of clients keep on increasing the available throughput across the network also drops substantially.
0
50
100
150
200
250
300
350
400
450
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19
Aggregate Throughput
MacBook pro Azuz Dongle Azuz Dongle Note 3
Mac (802.11n) Avg DELL (802.11n) Avg DELL (802.11g) Avg
ARUBA MANAGEMENT INTERFERENCE:
Aruba Adaptive Radio Management (ARM) technology optimizes Wi-‐Fi network behavior and automatically ensures that Aruba access points stay clear of RF interference, resulting in a more reliable, higher-‐performing wireless LAN. ARM uses patented technology that uses automatic infrastructure-‐based controls to manage the entire RF spectrum. ARM™ dynamically adjusts the RF environment to maximize Wi-‐Fi stability and predictability, ensuring optimal performance for all clients and applications.
Wi-‐Fi Challenges:
There are certain challenges in Wi-‐Fi environments. Wi-‐Fi being a shared medium, clients must compete for bandwidth and simultaneously avoid collisions. The problem of co-‐channel interference also creates a challenge in providing good service. An additional problem is that a slow client can make the whole WLAN slow and cause performance problems for every other client.
As Wi-‐Fi works in the unlicensed band, there is considerable interference caused by Bluetooth devices, cellular antennas, microwave ovens, wireless cameras. This interference can cause serious performance problems.
Another problem is that of sticky clients. This is caused by a fault in the devices, which connect, to an AP that is poor in performance and stays connected even though there are other APs available. This causes overload in APs, which in turn reduces the network and client performance.
As one of the groups worked on Wi-‐Fi Multimedia (WMM) it is important to address the challenges associated with it also. WMM provides QOS features that help in voice, video, and other latency-‐sensitive traffic get priority treatment over regular data. To provide the appropriate QOS and improve user experience it is necessary to address this.
Features of ARM:
1. Boost client performance: ARM works with Aruba ClientMatch technology to maximize client performance. ClientMatch continuously gathers session performance metrics from devices and uses this information to steer each one to the best radio on the best AP. ClientMatch dynamically optimizes Wi-‐Fi client performance as users roam and RF conditions change. If a device moves out of range of an AP or RF interference unexpectedly impedes performance, ClientMatch steers it to a better AP. ARM and ClientMatch technologies are critical to ensuring that overall network capacity and performance remains consistent.
2. App visibility and performance:
Aruba AppRF technology which is included in the ARM, h leverages information from Aruba Mobility Controllers to identify a wide range of business-‐critical enterprise applications and apply the appropriate QOS tags. ARM and AppRF ensure the web and mobile applications that are used in the enterprise get the end-‐to-‐end QOS tags that are needed. It also gives the IT department visibility and control over the traffic used in the voice and video traffic. ARM is the only RF management technology certified to optimize latency-‐sensitive applications. It dynamically adapts RF scanning and adjusts RF bandwidth available in the presence of mission-‐critical apps. Optimizes voice and video application performance and improves the user experience.
3. Ensure airtime fairness: Wi-‐Fi being a shared medium we need to ensure each client must be allotted fair amount of time on the air. ARM maximizes client performance by giving each client fair access and ensuring that no single client or group of clients monopolizes resources at the expense of others. This does not let one device to boss the entire network.
4. Reduce co-‐channel interference: Co-‐channel interference occurs when there are multiple devices trying to connect to the same channel simultaneously. This reduces performance of the channel and also reduces throughput for all the devices connected to that channel. To tackle this problem, ARM adjusts channels and transmit-‐power based on the changing the RF environment. For example, if an AP goes down, ARM automatically adjusts the transmit power of surrounding APs accordingly to fill coverage holes. ARM also coordinates access to a single channel, which allows neighboring APs to share the RF spectrum without increasing co-‐channel interference. Overcomes the challenges of dense AP deployments in the 2.4-‐GHz band.
5. Immune to non-‐Wi-‐Fi interference: ARM ensures that network access and data throughput is maintained, even in the presence of significant interference from non-‐Wi-‐Fi sources. ARM automatically adjusts the affected APs using a variety of techniques. This helps to ensure that the network continues to perform optimally. The techniques used are shrinking the coverage area, raising the noise floor, and throttling back the maximum throughput. Aruba APs also perform spectrum analysis. This enables APs to identify sources of non-‐Wi-‐Fi interference like video bridges, microwave ovens or Bluetooth devices.
6. Optimize spectrum usage: The number of channels in the 5-‐GHz band is limited and varies based on regulatory domain, PHY and RF coverage. Hence the network must adapt to these conditions and
use all active channels to maximize spectrum availability. ARM uses dynamic channel selection to automatically assign channel and power settings for all APs in the network. This method ensures that APs operate over the healthiest or least congested channels and the WLAN making efficient use of the available spectrum.
7. Channel scanning intervals: APs can go off-‐channel at fixed intervals to scan other channels for noise and rogue devices. This scanning helps APs choose a channel that maximizes their performance and is essential to detect unauthorized or malicious devices. ARM dynamically adjusts its channel scanning based on various parameters:
• AP load: ARM tracks the traffic load on each AP and dynamically adjusts the scanning frequency accordingly. A lightly loaded AP will scan frequently, such as once a second, while a heavily loaded AP will scan less often.
• Traffic type: ARM leverages information from Aruba Mobility Controllers to identify voice and video traffic traversing an AP. When these flows are detected, the AP stops scanning to avert latency. As a result, voice and video and the user experience are optimized.
8. Update neighboring APs: When an AP pauses scanning to accommodate application traffic, that AP still receives updates over the air from neighboring APs with information they’ve collected about the RF environment. Receiving these updates over the air enables the AP that has paused scanning to determine the signal strength and transmit power of the APs sending out updates the RF environment. This is important in creating a path-‐loss diagram and also mapping AP locations.
9. Adaptive power and channel assignments: Automatically assigns all AP channel and power settings. Supports 802.11n and 802.11ac wide-‐band channels, including 40 MHz and 80 MHz Automates many setup tasks during network installation and during ongoing operation when RF conditions change.
10. Mode-‐aware ARM: Aruba APs dynamically detect when radios have overlapping coverage, remove oversubscribed APs from serving clients, stop beaconing, and turn into air monitors. Provides additional network security through greater wireless intrusion detection and frees-‐up airspace for client traffic.
CONCLUSION:
From the above set of experiments and their results we observed the following important factors:
• Rate Verses Range: While an ideal test implementation would show very high throughput in terms of hundreds of Mbps at short range distances, these values will drop slowly at a few tens of feet and thereon will drop suddenly to almost no achievable throughput values. This is steep drop is decided by the received power level from the access point or the RSSI value of the form of -‐85 dBm. On the other hand when other clients are introduced in similar deployment the throughput decreases substantially due to channel sharing, also the range the decreases such that locations which are accessible in a one client scenario are then not accessible in case of multiple clients at same distances.
• Multi-‐Client Test: On one hand where a practical situation does not include just one host in the network, it was worth noting the way in which the throughput was almost lost when many clients were deployed in the same network which gave a very high throughput on one host implementation but lost such a performance when multiple clients were implemented. We not only observed that the throughput was decreased for each client, but also the aggregated throughput of the network dropped at a very high level, while new clients were added at regular intervals. We also observed an increase in performance when the clients were allowed to stable after adding all the test clients and then waiting for a sufficient amount of time.
We also tried to understand Aruba Radio Management interface, which is a smart solution to understand the variation in environment and adapt to changes such that the wireless network achieves its maximum performance.