Problem 3: Network Initialization

In the 802.16(d) mesh CDS mode, a new node is required to transmit its MSH-NENT messages on NENT TxOpps for requesting to join the network in a contention-based manner, while a node that has been operational²should transmit its MSH-NCFG messages on NCFG TxOpps for network maintenance and its DSCH messages on MSH-DSCH TxOpps for minislot scheduling in a collision-free manner (with the aid of MEA/TMEA-S/TMEA-D). Initializing an 802.16(d) mesh CDS-mode network using single-switched-beam antennas encounters four problems: 1) how a new node can receive MSH-NCFG messages transmitted by neighboring operational nodes to synchronize its clock and MAC-layer framing with those of the network; 2) when a new node can transmit its MSH-NENT messages so that its chosen sponsoring operational node can receive them; 3) how an operational node can receive an MSH-NENT message transmitted from a neighboring new node; and 4) when an operational node can transmit its MSH-NCFG messages so that neighboring new nodes can have chances to receive them. We answer these questions and briefly explain our refined network entry process for a new node using a single-switched-beam antenna below.

First, each operational node is required to periodically transmit its MSH-NCFG mes-sages to each of its TDs regardless of the inexistence of operational nodes in those TDs.

That is, the MEAIs of all the TDs of an operational node should be activated at all time after the node has joined the network. With this design, a new node can have a chance to receive MSH-NCFG messages of its neighboring operational nodes. For a new node, it should first scan all of its TDs to detect whether any operational node exists. This scanning procedure is described here. When a new node boots, it should iteratively point

2Recall that a node being operational means that it has joined the network and started transmitting network-management messages.

its antenna to each of its TD to try receiving any incoming messages. When pointing its antenna to a TD, the new node should continuously listen to this TD for 2⁸ NCFG TxOpps, unless it receives MSH-NCFG messages sent by the operational nodes in this TD. Because each operational node is allowed to use holdoff times ranged from 2⁰ to 2⁷, continuously listening to a TD for 2⁸ NCFG TxOpps is sufficient for a new node to receive an operational node’s MSH-NCFG messages, if any operational node exists in that TD. If the new node does not receive any messages in a TD, after listening for 2⁸ NCFG TxOpps, the new node should point its antenna to next TD and continue its scanning process. A new node should repeat the above process until it has scanned all of its TDs and detected the existence of neighboring operational nodes.

According to [1], a new node cannot transmit its own MSH-NENT message until it receives the MSH-NCFG messages sent from the same node twice. Thus, after detecting the existence of an operational node, the new node can stay in the TD where the detected operation node resides to wait for its next MSH-NCFG message. If multiple operational nodes in different TDs are detected, the new node can first wait for the MSH-NCFG messages transmitted by the operational node with the highest SINR value.

After receiving MSH-NCFG messages from the same node twice, the new node can choose one of its neighboring operational node as its sponsoring node and start its net-work entry process. In the netnet-work entry process, the sponsoring node is responsible for allocating a temporary bidirectional minislot allocation to relay network-joining control messages transmitted by the base station and the new node. This temporary minislot allo-cation is required because a new node has not been operational and thus cannot negotiate minislot allocations with any operational nodes.

In the network entry process, the new node is required to transmit several network-joining and registration control message to the base station. These control messages should be relayed via its chosen sponsoring node and be first sent to the sponsoring node using MSH-NENT messages on NENT TxOpps. To achieve this goal, two problems need to be solved: One problem is when a new node can transmit its MSH-NENT messages on certain NENT TxOpps so that its chosen sponsoring node can receive them. The other problem is how an operational node can receive such network-joining control messages from an unknown (new) node on NENT TxOpps. We first answer the second problem here: An operational node is required to iteratively point its antenna to each of its TDs

on NENT TxOpps. That is, on each NENT TxOpp, it points its antenna to a different TD. For example, a k-TD node has to point the antenna to its TD (i mod k) on NENT TxOpp i.

With the above arrangement, we describe the solution for the first problem here: For a new node, it first divides NENT TxOpps into several k-TxOpp groups. After choosing its sponsoring node, the new node should point its antenna to the TD where the sponsoring node resides on all NENT TxOpps. It then transmits its MSH-NENT messages that carry its network-joining control messages in a probabilistic manner. When the NENT TxOpp number advances to the boundary of a k-TxOpp group, a new node randomly chooses a fraction Nf between 0 and 1. If Nf is below a pre-determined value Nth, the new node is allowed to transmit its network-joining messages during this TxOpp group.

In this condition, it should continuously transmit its MSH-NENT messages on each of the k TxOpps in this TxOpp group. If Nf is larger than or equal to Nth, the new node should keep silent during this TxOpp group. Because an operational node sequentially switches its antenna to each of its k TDs on k TxOpps and a new node stays at one of its TD for k TxOpps, their antennas can meet each other on a certain TxOpp.

If the chosen sponsoring node receives an MSH-NENT message from a new node on NENT TxOpps, it should first add the ID of this new node into the eligible node list of each of its TD and considers the control message schedulings of this new node for MSH-NCFG and MSH-DSCH messages are in unknown statuses. That is, it should assume that the new node will contend for every MSH-NCFG and MSH-DSCH TxOpp; thus, its MEAIs should always take the contention of this new node for NCFG and DSCH TxOpps into consideration. In addition, the sponsoring node should disseminate the unknown schedule information of this new node to its neighboring nodes using its MSH-NCFG and MSH-DSCH messages, such that its neighboring nodes can also take the contention of this new node in consideration. The reason behind this design is that, if the nodes in the new node’s two-hop neighborhood do not consider its contention for NCFG and DSCH TxOpps in advance, after the new node becomes operational, the one-hop neighboring nodes of the new node will not point their antenna to receive the NCFG and MSH-DSCH messages transmitted by the new node and the two-hop neighboring nodes of the new node will not properly point their antennas to avoid inter-node scheduling conflicts.

The new node can know whether its chosen sponsoring node has received and accepted

its sponsorship request by checking whether the next MSH-NCFG message transmitted by the chosen sponsoring node contains the network-entry-open IE for acknowledging its sponsorship request. If the new node does not find the corresponding network-entry-open IE in the sponsoring node’s next MSH-NCFG message, it should halve the value of Nthand repeat the above process until the chosen sponsoring node responses it with a network-entry-open IE. The reason why a new node should decrease the value of Nth is to reduce the contention for NENT TxOpps, if multiple new nodes exist and simultaneously choose the same operational node as their sponsoring nodes. The lower bound of Nth is set to 0.1 in our simulations, which is sufficient for all new nodes to join the simulated network.

On the other hand, if the new node finds a network-entry-open IE for another new node in the next NCFG message of the chosen sponsoring node, it should cease its MSH-NENT message transmission to avoid disturbing the ongoing network entry process of another new node, until it finds that the chosen sponsoring node transmits an MSH-NCFG message containing a network-entry-ack IE for that new node, which indicates that that new node has completed its network entry process.

If the new node detects a network-entry-open IE for itself in the chosen sponsoring node’s MSH-NCFG message, it means that the sponsoring node has allocated a tempo-rary bidirectional minislot allocation for the new node to relay its network-joining and registration control messages. The new node should then transmit a network-entry-ack IE to its sponsoring node using MSH-NENT messages for acknowledging its sponsoring node (using the same way for transmitting its MSH-NENT message). After this, it can transmit its capacity negotiation, authorization, and network registration control mes-sages to the sponsoring node over the temporary minislots allocated by the sponsoring node for it. After receiving these control messages, the sponsoring node will in turn for-ward these messages tofor-wards the BS node in the network. Also, upon receiving control messages transmitted by the BS node and destined to the new node, the sponsoring node will forward them to the new node on the temporary minislots that it allocated for the new node.

After finishing exchanging necessary control messages with the BS node, the new node has joined the network and become operational. It should then transmit a network-entry-close IE to the sponsoring node using MSH-NENT messages on NENT TxOpps.

The network-entry-close IE is used to notify the sponsoring node that the network-entry

process of the new node that it sponsors has been completed and the temporary minislot allocation reserved for the new node is no longer needed. After receiving the network-entry-close IE, the sponsoring node first acknowledges the new node by sending it an MSH-NCFG message containing a network-entry-ack IE on NCFG TxOpps and then cancels the temporary minislot allocation reserved by it for the new node.

Notice that, for a node that just joined the network, it is only allowed to transmit its MSH-NCFG/MSH-DSCH messages in the TD where its chosen sponsoring node re-sides. The reason is that at this stage it is possible that only the nodes in the two-hop neighborhood of its chosen sponsoring node know the existence of this new operational node. If this new operational node transmits its control messages in other TDs, nodes in other TDs may not know on which TxOpps they should point their antennas toward the TDs where this new operational node is. Worse yet, they may schedule their own control message transmissions on the same TxOpps used by the new operational node, generating inter-node scheduling conflicts.

To solve this problem, we require a new operational node not to transmit its own control messages to TDs where nodes have not learned its existence. The new opera-tional node should continuously transmit MSH-NENT messages (without containing any network-joining messages) to each of its TDs (excluding the TD where its sponsoring node is) within each of the k-NENT-TxOpp groups in the probabilistic manner described above. Only when it finds the MSH-NCFG/MSH-DSCH messages transmitted from the same node in a TD contain its next NCFG/DSCH TxOpp information (indicating that this new operational node is in the unknown scheduling status and considered to contend for every NCFG/DSCH TxOpp) twice, can it stop transmitting its MSH-NENT message in this TD and start scheduling its next NCFG/DSCH TxOpp in this TD. This is be-cause now the new node can ensure that nodes in this TD know its existence and have notified their one-hop neighboring nodes of its existence. Thus, the transmissions of its MSH-NCFG/MSH-DSCH messages in this TD will not lead to any inter-node scheduling conflicts. If the new node does not receive any MSH-NCFG/MSH-DSCH messages in a TD after k² NENT TxOpps, it means that no operational node is present in this TD.

Thus, the new node can start scheduling its MSH-NCFG/MSH-DSCH message transmis-sions in this TD. Such control message transmistransmis-sions are still needed because there may be several new nodes in this TD.

3.2.5 Problem 4: Transmission-domain-aware Minislot