Comparison of WiFi adapters compatible with leJOS

As you know, the leJOS Java virtual machine, which we can use to program the LEGO Mindstorms EV3 robotic kit, is compatible not only with the officially supported NETGEAR N150 adapter, but also with a number of other adapters. Here I offer you a translation of an article from the official leJOS blog that compares the various supported WiFi adapters.

leJOS has always allowed different WiFi adapters, unlike the original Lego software that only supports Netgear devices. In the latest version of leJOS 0.9.0, the developers improved the existing drivers and added new ones. This means that we now have a wider selection of devices. In this article, I will compare 5 nano adapters with the original device. Let’s look at each device first.

	NETGEAR N150(WNA1100) This is an officially supported WiFi adapter. It uses the Atheros AR9271 chipset. Its cost is approximately £10. Works well, but much more so than the rest of the devices reviewed here. Must be selected if you want to use unmodified LEGO brand firmware.
	DIGITAZZ 150MBPS WIRELESS ADAPTOR This is a cheap “nano” adapter (but not as small as the others). Uses the Ralink RT 5370 chipset and costs approximately £5.
	The Pi Hut USB Wi-Fi Adapter for Raspberry Pi A “nano” adapter sold specifically for use with the rPi. Costs approximately £6. Uses Ralink RT5370 chipset.
	CSL – USB Wlan (WiFi) for PC / Raspberry Pi Another “nano” adapter aimed at the rPi market. Uses Realtek 8188cu chipset. Costs approximately £6.
	TP-LINK TL-WN725N 150Mbps Wireless-N Nano USB Adapter This “nano” adapter comes in two versions, but despite the use of different chipsets in these versions, they share the same model number. The second version tested here uses the Realtek 8188eu chipset. Costs approximately £6.
	Edimax EW-7811UN 150Mbps Wireless Nano This was the first adapter supported by leJOS. Uses Realtek 8188cus chipset and costs around £7.

Tests

To compare devices, I performed a couple of tests with each of them. First test is a simple ping test from my PC. During the test, I placed the EV3 (along with the adapter) around in different places. For most adapters, location had little to no effect on test results. However, the DIGITAZZ, Pi Hut and TP-Link adapters got into dead zones in some places. Such places were only a few meters from my WiFi router (A D-Link DIR-665). I have done this test many times with these adapters and the situation is easily repeated. I do not know why these three adapters have such a problem, perhaps the essence lies in the design of the antenna used or some kind of incompatibility with my router. In any case, this problem was not observed with the other four devices.

In the second test used a small program to simulate several different types of communication over a network. The program performed 4 tests:

- - A simple command send test that sends a 16-byte command packet and expects a 4-byte response.
  - A test to send a large amount of data, sending 1024 bytes of data and also expecting 4 bytes in response.
  - 1024 bytes streaming test.
  - Streaming test using large 64Kb packets.

In each case, the test was performed for 30 seconds and the data transfer rate was calculated. Each test was performed 3 times and the mean value is used as the result. All 4 tests were run in two directions (PC to EV3, EV3 to PC). Also, for comparison, I ran these same tests between a PC and an ultrabook. The results are shown below, all data in kbps:

	From PC to EV3				From EV3 to PC
Device	RPC	1024+Ack	1024	64K	RPC	1024+Ack	1024	64K
NETGEAR	6.57	178.43	642.70	643.38	6.04	160.29	591.87	593.90
Digitalzz	2.57	97.87	702.79	609.93	3.82	155.24	606.38	622.97
Pi Hut	1.40	45.37	767.75	727.73	1.37	51.54	574.47	472.36
CSL	4.12	141.56	555.44	546.88	4.34	92.76	676.79	680.77
TP-LINK	5.90	174.65	743.46	681.80	5.98	159.77	664.64	672.88
Edimax	4.27	131.47	725.84	708.73	4.11	93.05	685.80	687.25
Ultrabook	11.99	418.48	1665.01	1885.39	11.09	347.30	1552.45	1721.08

What’s better?

None of the tested adapters are perfect. The good news is that none of them are expensive and are offered at a reasonable price compared to a few years ago. Netgear provides good performance, but is too large to be used in mobile robots. Adapter TP Link fast, but the problem of dead zones calls into question its use. Pi Hut too slow for normal use when sending commands and receiving responses. Digitalzz slow and has the same deadzone problems as TP-Link. Remain adapters Edimax and CSLwhich have the same level of performance. Edimax slightly smaller and seems rather fragile, on both of my adapters part of the plastic casing has chipped off.

If the problem with dead zones could be solved, then, first of all, I would choose a TP-Link adapter, because. it is small, works well and seems well made. The choice between Edimax and CSL is not easy. But overall, the Edimax adapter has been very stable over many, many months of use, so I think I would choose this adapter.

Outcome

The test results are based on the use of an unreleased (at the time of testing) version of leJOS 0.9.0. Even though many of the listed devices can work with 0.9.0, I used updated (in the case of TP-Link completely new) drivers. My tests are based on using the WiFi router described above. Results may vary with other routers. As mentioned above, TP-Link devices come in two versions that use different chipsets but share the same name and model number. This is not at all uncommon, so if you buy one of these devices and it does not work, it means that the manufacturer may have replaced the chipset of the device. If you’re having problems with one of these devices (or want to share your results), please post on the leJOS forum.

Author: GLOOMYANDY
Translation: Alexey Valuev
The original article is here.

kids-toy