Changes between Version 1 and Version 2 of WARPLab/Examples/nodeSync

Timestamp:

May 13, 2013, 5:32:07 PM (11 years ago)

Author:

murphpo

Comment:

--

WARPLab/Examples/nodeSync

@@ -2 +2 @@
 
 This WARPLab example illustrates how to share clocks and triggers among WARP nodes to eliminate all timing and frequency offsets. This configuration has many uses. For example, when developing a new PHY it can be helpful to test performance with zero carrier frequency offset (CFO). This same configuration can also be used to synchronize multiple nodes for large-scale MIMO experiments, where the RF interfaces of multiple nodes are used together to form a meta node with many antennas.
+
+This example uses two nodes. One acts as the source for clocks and triggers. The other acts as the sink. The source/sink relationship only applies to the synchronization signals. Both nodes can transmit and receive normally.
 
 == Setup ==
@@ -22 +24 @@
 
 
+|| [[Image(nodeSync_hwCfg_source.jpg)]] || [[Image(nodeSync_hwCfg_sink.jpg)]] ||
+|| Source Node Configuration || Sink Node Configuration ||
 
 Once the hardware is connected and programmed you can run the example m code: [source:/ResearchApps/PHY/WARPLAB/WARPLab7/M_Code_Examples/wl_example_siso_txrx_nodeSync.m wl_example_siso_txrx_nodeSync.m].
 
+=== Observations ===
+The wl_example_siso_txrx_nodeSync script creates a figure with three plots:
+ * Real part of the transmitted signal
+ * Real part of the received signal 
+ * Phase difference (unwrapped) between the transmitted and received signals
+
+When you run the script using fully synchronized nodes, the plot will look like this:
+[[Image(nodeSync_constPhaseOffset.png)]]
+
+Note the constant phase difference between the Tx/Rx waveforms. If you re-run the script, this constant value will change. This is due to the RF transceivers being reset, which forces the PLLs to re-tune, leading to a new random phase offset between their RF carrier signals. If you skip the reset, the inter-radio phase offset will remain constant. You can observe this effect by commenting out the call to ''wl_initNodes()'' in the script (after running the script at least once).
+
+If you disable the sharing of clock signals between nodes (by de-asserting both CM-MMCX switches at the sink node and re-programming the FPGA), you can directly observe the carrier frequency offset between the RF transceivers. The resulting plot will show a constant slope for the Tx-to-Rx phase difference. The CFO is proportional to this slope. The figure below illustrates this.
+
+[[Image(nodeSync_CFO.png)]]