diff --git a/integration/service/oasis/stationgraph/doc.go b/integration/service/oasis/stationgraph/doc.go
index aa6454179ab..5a161eec2e1 100644
--- a/integration/service/oasis/stationgraph/doc.go
+++ b/integration/service/oasis/stationgraph/doc.go
@@ -1,12 +1,62 @@
-package stationgraph
-
 /*
-- Node represents charge station and wraps energy operation.  It could dected whether
-  a target node is reachable or not, or how far it could go after charging.
-- Neighbor records information from current node to oppsite one, such as distance,
-  duration, energy consumption, etc.
-- Graph represents charge station based node graph and provide functionality to find
-  best solution.  Graph will be constructed by sub-class, once it has been built then
-  orig and dest has been decided.  When sub-class construct the graph, it will crate
-  information about node, and each node will contains neighbors which could be explored.
+
+Package stationgraph builds charge station based graph and provide optimum charge solutions for given orig/dest.
+- To create station graph
+    + currEnergyLevel: current energy level for given electrical vehicle
+    + maxEnergyLevel: maximum energy capacity for given electrical vehicle
+    + strategy: Charge strategy be used for each charge station, it hides logic of different strategy of {time used for charging/wating, new energy got}.
+    + querier: topological querier which generates connectivity for building graph
+         * querier could be implemented by querying OSRM or pre-build conductivities
+         * querier must contains stationfindertype.OrigLocationID and stationfindertype.DestLocationID
+         * querier must provide ways to integrate stationfindertype.OrigLocationID and stationfindertype.DestLocationID into connectivity graph
+
+- station graph returns
+    + solution.Solution via GenerateChargeSolutions()
+    + station graph will call algorithm and built certain type of Graph to calculate result
+
+Data structure
+- Graph defines interface of a graph.
+- nodeGraph and mockGraph implements interface of graph
+- For mockGraph, it mainly used for algorithm testing.  For all the nodes, topological is hard coded at initial time.
+- For nodeGraph, which is built topological on the fly during running time.
+    + Its start and node need to be setted.
+    + physical node: 1-to-1 match to charge stations and orig/dest point.  For example, the following graph has 7 physical nodes
+    + logical node: for each charge station, we might exist charge station with different charge status, depend on time used in charge stations.  For example, the following graph has 17 logical nodes.
+         * given charge station 1, we have different choices
+         * station 1, charge to 60% of maximum energy then left
+         * station 1, charge to 80% of maximum energy then left
+         * station 1, charge to 100% of maximum energy then left
+    + connectivity: first built physical connectivity, then built logical
+         * physical connectivity is ansowered by interface of connectivitymap.Querier, eg, station 1 connects station 4 and station 5
+         * logical connectivity is build considering physical connection + final charging status
+         * logical connectivity's result is the final graph which algorithm runs on
+
+
+                    station 1
+               /       \    \
+              /         \    \
+             /          _\_____station 4
+            /          /  \     /        \
+           /          /    \   /          \
+start  -------   station 2  \ /           end
+           \          \     / \          /
+            \          \   /   \        /
+             \          \_/_____station 5
+              \          /     /
+               \        /     /
+                    station 3
+more information of this example graph could go to testion_graph_test
+
+
+Algorithm
+- dijkstra implements single direction dijkstra based on Graph's interface
+- Basically, classic graph algorithm could find path with minimum cost, several modifications with electric vehicle case:
+    + Whether two charge station is reachable or not.
+         * For example, we have a node charge to 60% of total energy, let's say maximum is 100 unit, so this node could reach other nodes which is reachable by 60 energy unit.
+    + Arrival energy.  For each node, we have a energy status when left previous charge station and will spend energy for reaching current node, arrival energy is calculated by that.
+    + Charging time needed.  Charging time depends on two things: arrival energy and the status of charging when you left
+         * For example, when reaching charge station, you have energy of 30, and you want to get 60% of maximum energy when left charge station for given node
+         * chargingstrategy.Strategy will evaluate the time needed for charging, which hides the logic of charging curve for specific vehicle.
+
 */
+package stationgraph