diff --git a/trust/trust-ensemble/overview.svg b/trust/trust-ensemble/overview.svg
new file mode 100644
index 0000000..e6e49a1
--- /dev/null
+++ b/trust/trust-ensemble/overview.svg
@@ -0,0 +1,479 @@
+
+
+
+
diff --git a/trust/trust-ensemble/writeup.typ b/trust/trust-ensemble/writeup.typ
new file mode 100644
index 0000000..d8a35d0
--- /dev/null
+++ b/trust/trust-ensemble/writeup.typ
@@ -0,0 +1,27 @@
+#align(center)[#text(weight: "bold", size:2em)[Computing Trust from a Committee of Sensors]]
+#v(3em)
+
+#figure(
+caption: "Overview of the data pipeline from the device under operation to the trust framework",
+image("overview.svg", width:100%),
+)
+
+Expanding on the previus work , the idea is to leverage the information comming from more than one sensor.
+The sensors can be of different types (side-channel, host-based, network-based, user-feedback, log-based, etc.), but they should all report a binary or ternary values.
+
+= Sensors
+
+#grid(
+columns: 3,
+inset: 10pt,
+fill: (x, y) =>
+ if calc.even(y) { luma(230) }
+ else { white },
+align: (left, right, right),
+[*Sensor*], [*Indicators*], [*Output*],
+[MAD + Policy], [High Consumption,Boots, Reboots, etc.], [Ternary],
+[BPV], [Boots, Abnormal Boot], [Binary],
+[File Scanner (ClamAV)], [Malware], [Binary/Ternary],
+[Log Scanner ], [Abnormal Activity, Malware, Abnormal Boots], [Binary]
+
+)