An RFID reader reads only one antenna at a time, so 28 separate read operations (inventories) are required to determine the situation seen by each antenna beam. The reader's reading speed depends significantly on the settings used and the size of the tag population. For example, if each antenna beam sees 100 tags, and one read operation takes about 800 ms, a Full Round will take 28 x 800 ms, or about 22 seconds. If an antenna beam doesn't see any tags, a read operation takes about 80 ms, and a Full Round without tags takes about 2.2 seconds.
The RAW data obtained from these 28 antenna beams is snapshot of a point in time. When two or more sets of RAW data from Full Rounds are available, the direction of tag movement can be inferred by comparing these snapshots. However, because a Full Round can take a relatively long time to complete, the resulting RAW data can become more like a blurred image.
Accuracy can be enhanced by reducing the Full Round time. For an in-and-out type application, you can minimize the number of antenna beams in use so the reader doesn't waste time scanning irrelevant beams for identification. For example, by using only beams 9 and 17 and 5 and 13, the Full Round time can be significantly shortened (300 ms to 3200 ms), resulting in a more accurate picture.
To better understand how this works, let's look at what RAW data contains. RAW data is the TagStorage sent by the InventoryStream. TagStorage contains all the tags the beam sees, including EPC, RSSI, beam number, and PhaseDiff values.
The following figure shows example RSSI values for a tag measured during one Full Round. The tag was about 2.5 meters away from the reader, with a 1.6 meters height difference.
Figure 1: Full Round RSSI values for a tag
In this example, the tag was so close to the BFA antenna that practically all its beams saw it. The only difference is the signal strength (RSSI) measured by different beams. The tag is closest to beam 27. If only beams 23+15 and 19+27 were used in the test, you can easily conclude that the tag is to the reader's right. When the tag eventually moves to the left, the changes in signal strength among the antenna beams allows the corresponding in or out event to be generated.
PhaseDiff indicates whether the tag is moving towards or away from the reader. Unfortunately, this value is usually very unstable, especially if the tag is read indirectly (for example, when a radio wave bounces through a wall to the tag). Therefore, only the PhaseDiff information measured by the strongest antenna should be used for direction detection.
Due to the thin structure of the BFA, each beam also has a corresponding ghost beam opposite it that is read at the same time, but with significantly less power. For example, beam 27's ghost beam is beam 23, and beam 11's is beam 7. Consequently, direction analysis cannot be performed until measurement results are obtained from both beams to determine which of them is the correct reading and which is the ghost reading to be ignored.