diff --git a/src/main/java/net/onrc/onos/ofcontroller/bgproute/IPatriciaTrie.java b/src/main/java/net/onrc/onos/ofcontroller/bgproute/IPatriciaTrie.java
new file mode 100644
index 0000000..9ec50bf
--- /dev/null
+++ b/src/main/java/net/onrc/onos/ofcontroller/bgproute/IPatriciaTrie.java
@@ -0,0 +1,20 @@
+package net.onrc.onos.ofcontroller.bgproute;
+
+import java.util.Iterator;
+
+public interface IPatriciaTrie {
+	public Rib put(Prefix p, Rib r);
+	
+	public Rib lookup(Prefix p);
+	
+	public Rib match(Prefix p);
+	
+	public boolean remove(Prefix p, Rib r);
+	
+	public Iterator<Entry> iterator();
+	
+	interface Entry {
+		public Prefix getPrefix();
+		public Rib getRib();
+	}
+}
diff --git a/src/main/java/net/onrc/onos/ofcontroller/bgproute/PatriciaTrie.java b/src/main/java/net/onrc/onos/ofcontroller/bgproute/PatriciaTrie.java
new file mode 100644
index 0000000..514758f
--- /dev/null
+++ b/src/main/java/net/onrc/onos/ofcontroller/bgproute/PatriciaTrie.java
@@ -0,0 +1,464 @@
+package net.onrc.onos.ofcontroller.bgproute;
+
+import java.util.Iterator;
+import java.util.NoSuchElementException;
+
+public class PatriciaTrie implements IPatriciaTrie{
+	private final byte maskBits[] = {(byte)0x00, (byte)0x80, (byte)0xc0, (byte)0xe0, (byte)0xf0, 
+												 (byte)0xf8, (byte)0xfc, (byte)0xfe, (byte)0xff};
+	
+	private int maxPrefixLength;
+	
+	private Node top;
+
+	public PatriciaTrie(int maxPrefixLength) {
+		this.maxPrefixLength = maxPrefixLength;
+	}
+
+	public synchronized Rib put(Prefix p, Rib r) {
+		if (p.getPrefixLength() > maxPrefixLength) {
+			throw new IllegalArgumentException(String.format(
+					"Prefix length %d is greater than max prefix length %d", 
+					p.getPrefixLength(), maxPrefixLength));
+		}
+		
+		if (p == null || r == null) {
+			throw new NullPointerException();
+		}
+		
+		Node node = top;
+		Node match = null;
+		
+		while (node != null
+				&& node.prefix.getPrefixLength() <= p.getPrefixLength()
+				&& key_match(node.prefix.getAddress(), node.prefix.getPrefixLength(), p.getAddress(), p.getPrefixLength()) == true) {
+		    if (node.prefix.getPrefixLength() == p.getPrefixLength()) {
+		    	/*
+		    	 * Prefix is already in tree. This may be an aggregate node, in which case
+		    	 * we are inserting a new prefix, or it could be an actual node, in which 
+		    	 * case we are inserting a new nexthop for the prefix and should return
+		    	 * the old nexthop.
+		    	 */
+		    	Rib oldRib = node.rib;
+		    	node.rib = r;
+		    	return oldRib;
+			}
+
+			match = node;
+			
+			if (bit_check(p.getAddress(), node.prefix.getPrefixLength()) == true) {
+				node = node.right;
+			} else {
+				node = node.left;
+			}
+		}
+
+		Node add = null;
+		
+		if (node == null) {
+			add = new Node(p, r);
+			
+			if (match != null) {
+				node_link(match, add);
+			} else {
+				top = add;
+			}
+		} else {
+			add = node_common(node, p.getAddress(), p.getPrefixLength());
+			if (add == null) {
+				//I think this is -ENOMEM?
+				//return null;
+			}				
+			
+			if (match != null) {
+				node_link(match, add);
+			} else {
+				top = add;
+			}
+			node_link(add, node);
+			
+			if (add.prefix.getPrefixLength() != p.getPrefixLength()) {
+				match = add;
+				
+				add = new Node(p, r);
+				node_link(match, add);
+			}
+			else {
+				add.rib = r;
+			}
+		}
+		
+		//If we added a new Node, there was no previous mapping
+		return null;
+		//return addReference(add);
+	}
+	
+	/*exact match*/
+	public synchronized Rib lookup(Prefix p) {
+		//TODO
+		
+		if (p.getPrefixLength() > maxPrefixLength) {
+			return null;
+		}
+		
+		/*
+		Node node = top;
+		
+		while (node != null
+				&& node.prefix.getPrefixLength() <= p.getPrefixLength()
+				&& key_match(node.prefix.getAddress(), node.prefix.getPrefixLength(), p.getAddress(), p.getPrefixLength()) == true) {
+			if (node.prefix.getPrefixLength() == p.getPrefixLength()) {
+				//return addReference(node);
+				return node.rib;
+			}
+			
+			if (bit_check(p.getAddress(), node.prefix.getPrefixLength()) == true) {
+				node = node.right;
+			} else {
+				node = node.left;
+			}
+		}
+		*/
+		
+		Node node = findNode(p);
+		
+		return node == null ? null : node.rib;
+	}
+	
+	/*closest containing prefix*/
+	public synchronized Rib match(Prefix p) {
+		//TODO
+		return null;
+	}
+	
+	public synchronized boolean remove(Prefix p, Rib r) {
+		Node child;
+		Node parent;
+		
+		if (p == null || r == null) {
+			return false;
+		}
+		
+		Node node = findNode(p);
+		
+		if (node == null || node.rib == null || !node.rib.equals(r)) {
+			//Given <prefix, nexthop> mapping is not in the tree
+			return false;
+		}
+		
+		if (node.left != null && node.right != null) {
+			//Remove the Rib entry and leave this node as an aggregate node
+			//In the future, maybe we should re-evaluate what the aggregate prefix should be?
+			//It shouldn't necessarily stay the same.
+			//More complicated if the above prefix is also aggregate.
+			node.rib = null;
+			return true;
+		}
+		
+		if (node.left != null) {
+			child = node.left;
+		} else {
+			child = node.right;
+		}
+		
+		parent = node.parent;
+		
+		if (child != null) {
+			child.parent = parent;
+		}
+		
+		if (parent != null) {
+			if (parent.left == node) {
+				parent.left = child;
+			} else {
+				parent.right = child;
+			}
+		} else {
+			top = child;
+		}
+		
+		/*
+		 * TODO not sure what to do here. I think this is lazily deleting aggregate nodes,
+		 * notice that it used to do nothing if it detected both children were not null earlier.
+		 * But here, what we really should do is reevaluate the aggregate prefix of the parent
+		 * node (if it is indeed an aggregate). Because at the moment, no aggregate node will ever
+		 * be removed. BUT, I don't actually think this presents a correctness problem, at
+		 * least from an external point of view.
+		 */
+		//if (parent != null && parent.refCount == 0) {
+			//node_remove(parent);
+		//}
+		
+		return true;
+	}
+	
+	public Iterator<Entry> iterator() {
+		return new PatriciaTrieIterator(top);
+	}
+	
+	private Node findNode(Prefix p) {
+		Node node = top;
+		
+		while (node != null
+				&& node.prefix.getPrefixLength() <= p.getPrefixLength()
+				&& key_match(node.prefix.getAddress(), node.prefix.getPrefixLength(), p.getAddress(), p.getPrefixLength()) == true) {
+			if (node.prefix.getPrefixLength() == p.getPrefixLength()) {
+				//return addReference(node);
+				return node;
+			}
+			
+			if (bit_check(p.getAddress(), node.prefix.getPrefixLength()) == true) {
+				node = node.right;
+			} else {
+				node = node.left;
+			}
+		}
+		
+		return null;
+	}
+	
+	/*
+	 * Receives a 1-based bit index
+	 * Returns a 1-based byte index
+	 * eg. (0 => 1), 1 => 1, 8 => 1, 9 => 2, 17 => 3
+	 */
+	private int getByteContainingBit(int bitNumber) {
+		return Math.max((bitNumber + 7) / 8, 1);
+	}
+	
+	private boolean key_match(byte [] key1, int key1_len, byte [] key2, int key2_len) {
+		//int offset;
+		//int shift;
+		
+		if (key1_len > key2_len) {
+			return false;
+		}
+		
+		int offset = (Math.min(key1_len, key2_len)) / 8;
+		int shift = (Math.min(key1_len, key2_len)) % 8;
+		
+		if (shift != 0) {
+			if ((maskBits[shift] & (key1[offset] ^ key2[offset])) != 0) {
+				return false;
+			}
+		}
+		
+		while (offset != 0) {
+			offset--;
+			if (key1[offset] != key2[offset]) {
+				return false;
+			}
+		}
+		return true;
+	}
+	
+	private boolean bit_check(byte [] key, int key_bits) {
+		int offset = key_bits / 8;
+		int shift = 7 - (key_bits % 8);
+		int bit = key[offset] & 0xff;
+
+		bit >>= shift;
+		
+		if ((bit & 1) == 1) {
+			return true;
+		} else {
+			return false;
+		}
+	}
+	
+	private void node_link(Node node, Node add) {
+		boolean bit = bit_check(add.prefix.getAddress(), node.prefix.getPrefixLength());
+		
+		if (bit == true) {
+			node.right = add;
+		} else {
+			node.left = add;
+		}
+		add.parent = node;
+	}
+	
+    private Node node_common(Node node, byte [] key, int key_bits) {
+		int i;
+		int limit = Math.min(node.prefix.getPrefixLength(), key_bits) / 8;
+
+		for (i = 0; i < limit; i++) {
+			if (node.prefix.getAddress()[i] != key[i]) {
+				break;
+			}
+		}
+		
+		int common_len = i * 8;
+		int boundary = 0;
+
+		if (common_len != key_bits) {
+			byte diff = (byte)(node.prefix.getAddress()[i] ^ key[i]);
+			byte mask = (byte)0x80;
+			int shift_mask = 0;
+			
+			while (common_len < key_bits && ((mask & diff) == 0)) {
+				boundary = 1;
+
+				shift_mask = (mask & 0xff);
+				shift_mask >>= 1;
+				mask = (byte)shift_mask;
+
+				common_len++;
+			}
+		}
+		
+		//Node add = new Node(null, common_len, maxKeyOctets);
+		//if (add == null)
+			//Another -ENOMEM;
+			//return null;
+		
+		//Creating a new Prefix with a prefix length of common_len
+		//Bits are copied from node's up until the common_len'th bit
+		//Rib is null, because this is an aggregate prefix - it's not
+		//actually been added to the trie.
+		
+		byte[] newPrefix = new byte[getByteContainingBit(maxPrefixLength)];
+		
+		int j;
+		for (j = 0; j < i; j++)
+			newPrefix[j] = node.prefix.getAddress()[j];
+
+		if (boundary != 0)
+			newPrefix[j] = (byte)(node.prefix.getAddress()[j] & maskBits[common_len % 8]);
+		
+		return new Node(new Prefix(newPrefix, common_len), null);
+		//return add;
+	}
+	
+	private class Node {
+		public Node parent = null;
+		public Node left = null;
+		public Node right = null;
+		
+		public Prefix prefix;
+		public Rib rib;
+		
+		public Node(Prefix p, Rib r) {
+			this.prefix = p;
+			this.rib = r;
+		}
+		
+		public Entry getEntry() {
+			return new PatriciaTrieEntry(prefix, rib);
+		}
+	}
+	
+	private class PatriciaTrieEntry implements Entry {
+		private Prefix prefix;
+		private Rib rib;
+		
+		public PatriciaTrieEntry(Prefix prefix, Rib rib) {
+			this.prefix = prefix;
+			this.rib = rib;
+		}
+		
+		@Override
+		public Prefix getPrefix() {
+			return prefix;
+		}
+		
+		@Override
+		public Rib getRib() {
+			return rib;
+		}
+	}
+	
+	private class PatriciaTrieIterator implements Iterator<Entry> {
+		
+		private Node current;
+		private boolean started = false;
+		
+		public PatriciaTrieIterator(Node start) {
+			current = start;
+			
+			//If the start is an aggregate node fast forward to find the next valid node
+			if (current != null && current.rib == null) {
+				current = findNext(current);
+			}
+		}
+
+		@Override
+		public boolean hasNext() {
+			if (current == null) {
+				return false;
+			}
+			
+			if (!started) {
+				return true;
+			}
+			
+			return findNext(current) != null;
+		}
+
+		@Override
+		public Entry next() {
+			if (current == null) {
+				throw new NoSuchElementException();
+			}
+			
+			if (!started) {
+				started = true;
+				return current.getEntry();
+			}
+			
+			current = findNext(current);
+			if (current == null) {
+				throw new NoSuchElementException();
+			}
+			
+			return current.getEntry();
+		}
+
+		@Override
+		public void remove() {
+			// TODO This could be implemented, if it were needed
+			throw new NoSuchElementException();
+		}
+		
+		private Node findNext(Node node) {
+			Node next = null;
+			
+			if (node.left != null) {
+				next = node.left;
+				//addReference(next);
+				//delReference(node);
+				//return next;
+			}
+			else if (node.right != null) {
+				next = node.right;
+				//addReference(next);
+				//delReference(node);
+				//return next;
+			}
+			else {
+				//Node start = node;
+				while (node.parent != null) {
+					if (node.parent.left == node && node.parent.right != null) {
+						next = node.parent.right;
+						//addReference(next);
+						//delReference(start);
+						//return next;
+						break;
+					}
+					node = node.parent;
+				}
+			}
+			
+			if (next == null) {
+				return null;
+			}
+			
+			//If the node doesn't have a rib, it's not an actual node, it's an artifically
+			//inserted aggregate node. We don't want to return these to the user.
+			if (next.rib == null) {
+				return findNext(next);
+			}
+			
+			return next;
+		}
+	}
+}