MoCA 2.0—Next-gen benefits and enhancements with backward compatibility—Part I

Pay TV service providers globally are delivering new services and applications enabled by the deployment of their managed home networks. These home networks are unique in that they require technology that can provide large throughput with high reliability, coverage to every TV, and coexist with legacy services. The technology must be capable of distributing multiple streams of HD video and broadband Internet data service throughout a subscriber’s home. It also must be extremely robust, able to deliver high-quality service and adapt to an almost limitless variety of environments depending on the layout of a subscriber’s home.

Increasingly, more operators are choosing Multimedia over Coax Alliance (MoCA), the only home networking standard designed from the ground up to support streaming video data over coax cables already in place. This article focuses on MoCA’s next generation standard, version 2.0 and the technical changes and benefits that advance the technology even further, yet remain backward compatible to the millions of homes already using MoCA.

Today, MoCA versions 1.0/1.1 are widely used by North American pay TV operators to offer multi-room digital video recorder (DVR) service, IP based video-on-demand (VOD) and bring broadband connectivity to set-top boxes and networked televisions, that enable a vast array of software applications that can be interactively run by subscribers. Verizon, with its FIOS data/TV/voice service uses MoCA extensively to deliver broadband data service and VOD. Such cable operators as Comcast, Time Warner, Cox and Charter deploy set-top boxes (STB) and digital-video-recorders with MoCA to enable multi-room DVR. Satellite TV operator DIRECTV features MoCA in its latest generation of STBs and DVRs and uses DIRECTV Ethernet to Coax Adapters (DECA) to attach to the broadband router and move older equipment onto the network.

Dish Networks also recently announced their MoCA enabled STB, DVR and MoCA adapters. By using a high-speed MoCA network to connect DVRs, STBs and broadband access points, the satellite TV operators can now offer multi-room DVR and use the Internet to access thousands of titles for streaming VOD. In other parts of the world such as Europe, MoCA is gaining traction as service operators conduct technology trials and find that as their bandwidth needs grow only MoCA has the ability to serve today’s requirements more reliably than any other home network technology as well as roadmap to adequately serve tomorrow with next-generation MoCA 2.0.

MoCA 2.0 more than doubles the available throughput of MoCA 1.1. Providing a minimum of 400 Mbps of useable MAC throughput in its baseline profile. MoCA 2.0 is able to keep pace and distribute more video and even greater levels of broadband service brought about by DOCSIS 3.0 and fiber-to-the-home (FTTH) deployments. If more than 400 Mbps is desired there are profiles for turbo modes and channel-bonded modes to enable upwards of 1 Gbps of MAC throughput. MoCA 2.0 offers the same outlet coverage operators have come to expect with MoCA 1.1 with increased reliability using new forward error correction for lower packet error rates. Also new for MoCA 2.0 are increased MAC efficiency, AES based privacy keys, network power management and enhanced parameterized quality-of-service (QoS)--all with backward interoperability with existing MoCA 1.0/1.1 products.

Requirements for Home Networking Over Coax

Channel Characteristics and Coexistence on the Coax

MoCA was designed to transport high-speed streaming video data over the existing coax cable distribution plants in homes.  Coax cable is the ideal medium as television displays are connected to the coax distribution system and the coax cable is shielded and capable of carrying high bandwidth signals.  The challenge of sending high-speed data over coax is that the distribution plants are not designed for room-to-room signal flow but are designed to feed a common signal to different rooms in the house.  Coax distribution plants are typically designed with a trunk and branch topology where the coax outlet in each room is connected to the trunk through splitters and individual pieces of coax cable.  The consequences of this topology are that room-to-room channel characteristics are not well controlled and may include severe multipath and high path loss.  The MoCA system deals with these channels by using bit-loaded OFDM to adapt to each channel independently and by transmitting high signal power.

In order to operate on existing coax in the home with minimum disruption, MoCA was also designed to coexist with other signals on the coax.  MoCA is capable of operating on various frequencies to avoid existing signals.  For example, in a home with CATV signals below 1 GHz, MoCA operates above 1125 MHz in a band called D. In a home with satellite L-band signals above 950 MHz MoCA operates between 475 and 625 MHz in a band called E.  To further minimize interference to existing services, MoCA also features transmit power control (TPC).  TPC reduces the MoCA transmit power by up to 30 dB.  Reducing transmit power lowers the likelihood that the MoCA signal will cause interference to devices operating in other bands.

Packet Type and Routing

The MoCA network accepts and transparently bridges Ethernet packets.  By providing support for Ethernet packets, MoCA can seamlessly integrate into any home data network.  All data including video and audio must be encapsulated in Ethernet packets before being transported across the MoCA network.  MoCA 1.x nodes use the MAC Source Address (SA), MAC Destination Address (DA) to route packets between MoCA nodes, and the VLAN priority (802.1p) to prioritize the transmission of the packets across the MoCA network.  MoCA 2.0 adds the option to use the DSCP field in the IP header (if present) for prioritizing the packet transmissions.  Because the MoCA network is not a broadcast network and packets are unicast between MoCA nodes, MoCA nodes cannot act as learning bridges by snooping traffic.  Instead each MoCA node monitors the SA of packets entering the MoCA network through it and distributes that information to all the other nodes via notification messages.  The notification messages inform other nodes that any messages destined to that address should be forwarded to the notifying node.

Throughput, Latency, Packet Loss

The MoCA system was designed to deliver “no excuses” streaming video in the home.  To meet this goal, MoCA targeted high throughput, low latency, and low packet loss.  High throughput was required to accommodate transporting multiple simultaneous HDTV streams.  A single HDTV stream typically ranges anywhere from 4 to 20 Mbps, while trick modes (e.g. fast forward or reverse) might peak at 2 to 4 times higher rates.  A home with four HD streams could consume 80 Mbps plus even more for trick modes.   A 16-node MoCA 1.1 network can deliver more than 140 Mbps of network throughput and MoCA 2.0 increases this to over 400 Mbps.

Low latency is important for three reasons, first to minimize video delay and buffering, second to allow digital rights management (DRM) schemes such as DTCP-IP to work, and third to maximize TCP-IP performance.  MoCA 1.1 has a maximum latency of 7.5 ms. Besides introducing delay, latency adds cost for buffer memory since packets “in-flight” across the network must be buffered in data buffers within the MoCA network.  To accommodate a 7.5 ms delay, the MoCA devices need to buffer up to 7.5 ms worth of data--at 100 Mbps a 750 Kbit buffer is required.  As delay and throughput increases, the size and cost of the buffers increase. 

DTCP-IP is one of the digital rights management (DRM) schemes adopted by the Digital Living Network Alliance (DLNA).  To operate, DTCP-IP monitors round-trip latency between the server and client to ensure they are both on the same local network and not separated by the Internet cloud.  If latency is too large DTCP-IP will shut down the flow.  Hence low latency is required to support DLNA networks and is also needed for the best TCP-IP performance.  It is well known that TCP-IP throughput is inversely proportional to the round-trip delay1 so as delay increases, TCP-IP slows down.  Tests have shown that latency in MoCA 1.x can limit TCP-IP performance to less than 100 Mbps.  One of the design goals for MoCA 2.0 was to improve TCP-IP performance by reducing latency.  As shown in Table 1, the average latency for a MoCA 2.0 network was reduced to 3.6 ms compared to 4.5 ms for a MoCA 1.1 network.

For streaming video (i.e. IP multicast), where there is no data retransmission, low packet loss is required to deliver high quality glitch free video.  MoCA 1.x guarantees a packet loss rate of less than 1e-5, and MoCA 2.0 lowered the allowed packet loss rate to less than 1e-6 or 1e-8 selectable by the higher layers. 

Security

MoCA provides a privacy feature.  When it is turned on, all data packets are encrypted.  MoCA 1.x uses 56 bit DES encryption while MoCA 2.0 upgraded the encryption to 128 bit AES.  Every MoCA device has a programmable password.  Devices that share the same password will form a network with each other while devices with different passwords will avoid each other.  Passwords are useful for distinguishing between networks in adjacent homes where the MoCA signal can leak between homes over a cable operator’s network.

Quality of Service

In addition to streaming video, the MoCA network may be used to simultaneously transport other services such as best effort Internet traffic, file transfers, etc.   When this happens, the video stream will contend for bandwidth on the MoCA network, which may result in dropped video packets.  MoCA offers two methods to protect the video traffic from other traffic.  The first method is to send video as prioritized traffic.  MoCA inspects packets that have VLAN tags and prioritizes traffic based on the three priority bits in the VLAN tag.  By sending video with high VLAN priorities, MoCA will give video precedence over untagged and lower priority packets and eliminate any contention between them.  The second method to protect the video is to send video using parameterize quality of service (pQoS).  Within MoCA, pQoS flows with specific flow parameters can be admitted to the network.  Once admitted, these flows are guaranteed the bandwidth required to be transported across the network.  pQoS flows can be uniquely identified by the content of their packet headers.  Packets classified as belonging to a pQoS flow are assigned the highest priority and their delivery is guaranteed.

MoCA 1.x Technical Backgrounder

Table 1 lists basic parameters of MoCA 1.0, MoCA 1.1, and MoCA 2.0.   This section provides more information about MoCA 1.x while the next section will explain changes made for MoCA 2.0. 

MoCA 1.x PHY Layer

MoCA 1.0 and MoCA 1.1 PHY layers are identical.  Both use bit-loaded OFDM with 224 subcarriers in a 50 MHz channel.   Bit-loaded OFDM was selected for MoCA because it is robust against multipath and optimizes the modulation between every pair of devices.  The MoCA 1.0 field trial showed that this technique has a 95% chance of delivering 100 Mbps between any two rooms in a home without any remediation.  This translates to 140 Mbps for MoCA 1.1.  Bitloading is performed periodically between all the nodes to track slow changes in hardware or the cable distribution plant due to temperature or other effects.  When bit-loading, each MoCA device probes the channel between itself and every other MoCA device in the network and selects the modulation on each of the 224 subcarriers based on the probe results:  the better the SNR on a subcarrier, the higher the modulation assigned to that subcarrier.  MoCA 1.x uses a maximum subcarrier modulation of 256 QAM.  Since the MoCA PHY layer adapts each link between node pairs independently, the channel capacity can be different between different nodes as well as between the forward and reverse directions of the same nodes.  The bit-loading parameters for a particular path are called a PHY Profile. 

Figure 1 illustrates how the MoCA network is composed of a collection of individual PHY Profiles, which can have different channel capacities.  Whenever data is sent between two MoCA nodes, the PHY Profile between those two nodes is used.  In addition to unicast PHY Profiles between node pairs, MoCA nodes also create a broadcast PHY Profile between each node and all the other nodes in the network.  The broadcast PHY profile is used to transmit Broadcast and Multicast data and is not as efficient as unicast PHY Profiles.  

MoCA PHY layer packets are sent as shown in Figure 2.  Packets start with an inter-frame gap (IFG) where nothing is sent.  The IFG separates the packet from the previous packet.  The IFG is followed by a preamble, which is used to delineate the start of the packet.  Channel Estimation (CE) symbols are then sent followed by data symbols used by the receiver to recover data.  Finally the packet ends with a cyclic redundancy checksum (CRC).  MoCA 1.x packets include Reed-Solomon parity bits in the data symbols to help keep packet loss below 1e-5.

Since MoCA shares the coax with other signals, transmit power control (TPC) is used between MoCA nodes to minimize interference to other devices using the coax.  During each bitloading cycle, the signal power between nodes is adjusted along with the modulation. TPC can reduce transmit power by up to 30 dB. 

The MoCA PHY layer can also operate at different frequencies.  Table 2 shows all the center frequencies, which can be used by MoCA 1.x devices.  The actual MoCA 1.x channel occupies 50 MHz.  D-band is the band to use when sharing the coax with a cable operator signal and E-band is the band to use when sharing the coax with a satellite signal.  Bands A and B are currently not used and C-band is used in Verizon FiOS homes.

The MoCA 1.x MAC Layer

The MoCA MAC uses a time division duplex (TDD) scheme where all nodes transmit on the same frequency during different time slots that are coordinated by a single node called the Network Coordinator (NC).  The NC is dynamically selected from all the nodes in a network based on the best broadcast bit-loading.  To ensure all nodes operate on the same time clock, all the nodes synchronize their clocks to the NC’s time that is broadcast to all the nodes approximately every millisecond in a MAP message.  MAP messages also carry dynamic scheduling information, which defines when each node can transmit in an upcoming time period called a MAP Cycle.  By specifying how every time slot is allocated, the NC ensures no contention or collisions occur on the network, which is necessary to support high reliability needed for streaming video.

During each MAP Cycle, some MoCA nodes are given the opportunity to send a reservation request (RR) message to the NC whereby the  nodes request time slots to send messages to other nodes.  The NC responds to all the RRs it receives in a MAP Cycle by granting time in the next MAP Cycle for as many transmissions it can, and sending a schedule for those grants in the next MAP message.  When nodes receive the schedule in the MAP message, they know when they should send and receive data during the upcoming MAP Cycle.  Nodes that have received Ethernet packets from an upper layer to send across the MoCA network are called ingress nodes.  In the RR requests, ingress nodes include priority information and MoCA 1.1 nodes also include pQoS information.  The NC uses this information to schedule pQoS data first followed by prioritized data in priority order followed by best effort data. 

Since the NC sees the priorities of all the packets in the network, it is able to prioritize packets on a network-wide basis and not just locally.  If nodes request more data than the network can carry, best effort and lower priority packets are dropped first from the schedule.

Table 1 shows MoCA 1.x data throughput declines as a MoCA network goes from two nodes to more nodes.  The decline is a direct result of the overhead associated with granting additional nodes opportunities to send RRs.  As new nodes are added to a network, the NC must schedule additional RRs, which take away transmission time slots from data.  The NC will add additional RRs per MAP cycle for each node up to six nodes, including the NC, after this the NC will use “multi-cycle” scheduling where nodes do not get an RR grant every MAP cycle.  Instead they may only get an RR opportunity every two or three MAP cycles.

Latency in the MoCA network is also dominated by the request-grant process.  From the time an Ethernet packet arrives at an ingress node, to the time the node gets a grant time to send the packet, several milliseconds may have passed. 

Total latency is comprised of the following times:

1. Packet arrival to RR time:  The time between the arrival of an Ethernet packet at the ingress node till the time the node can transmit a RR which requests a transmission slot for that packet.  This may take several MAP cycles.

2. RR to MAP time:  The time between an ingress node sending a RR and receiving the next MAP, which grants time slots to transmit.

3. MAP to grant time:  The time from receiving the MAP till the actual transmission time granted in the MAP

4. Egress time:  The time it takes for the receiving node (Egress Node) to forward the packet after receiving the packet on the coax

Although MoCA 1.0 and MoCA 1.1 use the same request-grant mechanism and the same MAP cycle duration, a MoCA 1.1 network can have more latency than a MoCA 1.0 network. Table 1 shows that MoCA 1.0 has less latency than MoCA 1.1.  The reason for this difference is because a MoCA 1.1 network can have 16 nodes while a MoCA 1.0 network is limited to 8 nodes.  In MoCA 1.x the NC typically grants 5 or less RR opportunities during each MAP Cycle so when there are more than six nodes in the network (including the NC), a node must wait longer between RR opportunities.  So since there may be more nodes in a MoCA 1.1 network, the MoCA 1.1 nodes may experience longer packet arrival to RR times.   Latency in a MoCA 1.1 network with 8 nodes or less will be the same as latency in a MoCA 1.0 network with the same number of nodes.

To achieve higher throughput, MoCA 1.1 adds packet aggregation to MoCA 1.0.  Packets from an ingress node to the same egress node can be aggregated together and sent in one PHY layer packet (Figure 2).  By aggregating packets the PHY layer overhead is shared and MoCA 1.1 can get a maximum throughput of 176 Mbps compared to 125 Mbps for MoCA 1.0.  MoCA 1.1 allows aggregation of up to 6 KB into one PHY layer packet.

Next:  MoCA 2.0

About the Author

Ron Lee is Strategic Technology Director at Entropic Communications. He can be reached at: ron.lee@entropic.com