We have covered the home networking battle between HomePlug and G.hn in various articles before. Peter White has an excellent piece analyzing the positions of the respective camps here. G.hn is quite an attractive proposition, aiming to get a networking device work over multiple media (coax, phone lines and power lines). As such, it has the ability to fully replace MoCA, HomePNA and HomePlug in the networking space. G.hn is an ITU standard which was ratified in June 2010.

While the G.hn consortium was working on the standardization and getting silicon ready, HomePlug has quietly been expanding its install base. It is finally the extent of market penetration which will decide the winning standard. Will service providers be ready to start afresh for technology / silicon which might currently appear superior on paper? While G.hn compliant silicon can have a PHY rate of 1 Gbps, the IEEE P1901 compliant Qualcomm AR7400 has a PHY rate of 500 Mbps. The upcoming HomePlug AV2 (HPAV2) standard is expected to match G.hn in terms of PHY rates.

G.hn Silicon

Sigma Designs took the early lead in introducing G.hn compliant silicon with the CG5110. We saw this chip in action at CES 2011. Lantiq also demonstrated their G.hn chipset at the same time. Today, Marvell is announcing the availability of their first G.hn chipset (the 88LX3142 digital baseband processor transceiver and the 88LX2718 baseband analog front-end) along with a reference design and firmware (SPIRIT) based on the UPA PLC (United Powerline Alliance's powerline communication device).

Marvell's presence in the G.hn market can be traced to its acquisition of the Spanish powerline networking firm, DS2 (Design of Systems on Silicon). DS2's powerline standards organization, the UPA, had been up against HomePlug till 2009. However, it could only strike base with some service providers like Telefonica and SinTel, and could never get into the retail market. The presence of HomePlug products in the retail market has served to extend its install base, making system manufacturers and also the relevant service providers hesitant to move on to non-interoperable, or worse, non-co-existing standards.

Marvell indicated that more than 20 early adopters (including OEMs and ODMs) have opted for Marvell's silicon and it probably includes a major chunk of the erstwhile UPA's customers. The fact that the 88LX3142 can coexist with the UPA installations should have made it easier for them. Retail availability of products based on this silicon is expected to be towards the end of Q4 2012.

G.hn's support for multiple media is also envisaged as the target for the IEEE P1905 standard. G.hn proponents cite P1905's drawbacks with respect to system and software complexity when dealing with systems talking to multiple media. IEEE P1905 proponents indicate the extra cost burden incurred by G.hn customers when dealing with devices which don't need to talk to multiple media.

Consider the usecase of a connected TV, where the antenna signal comes through a coax cable. In that case, the TV can connect to the home network via powerline.  However, the coax is not active all the time. In this case, the TV can shift to the coax for networking because it would deliver higher throughput / QoS. With G.hn, this can be done with a single chip, but IEEE P1905 would require complex software and product design as shown below.

The use case definitely appears very interesting, and it is good that we are having multiple vendors coming out with silicon around G.hn. However, it takes more than a year to move silicon to deployment (especially with service providers). In the meanwhile, HomePlug / P1901 devices are expanding their base. This brings us to the next aspect, namely interoperability and coexistence.

Irrespective of the technical capabilities of the products based on the two standards, it is essential that the adopted standard not disrupt existing installations of products based on other standards. This is especially important for MDUs (Multiple Dwelling Units), where the power lines are shared across multiple residences. One would definitely not want a G.hn device to mess up the operation of a HomePlug device. Considering that HomePlug already has an installed base, and there is no G.hn silicon with consumers yet, we believe that G.hn vendors must ensure that they peacefully coexist with them when it comes to the retail market. The G.9972 component of the standard specifies a coexistence mechanism for G.hn devices.

IEEE P1901 specifies a mandatory ISP (indicated with a star in the above timeline). This Inter-System-Protocol (also termed as Inter-Specification-Protocol in some documents) ensures that a IEEE P1901 device can coexist (with just loss in throughput) with a G.hn device. One must note that the G.hn device must also support the ISP (specified as G.cx). The chipset introduced by Marvell supports G.cx and it should be able to co-exist with IEEE P1901 devices implementing the ISP segment of the specification. In July 2011, Kawasaki Microelectronics (part of the HD-PLC alliance) introduced silicon compliant with IEEE P1901 and also implementing the ISP. Marvell indicated that the 88LX2142 could definitely coexist with such HD-PLC solutions. Note that coexistence implies absence of interference only. It doesn't mean that a UPA device can talk with a device based on G.hn silicon from Marvell.

In the above timeline, I have specifically refrained from indicating the availability of IEEE P1901 compliant HomePlug silicon (despite the fact that many chipsets such as the AR7400 from Qualcomm claim to be IEEE P1901 compliant). We have not found any documentation pointing to the existence of the mandatory ISP in it. (The AR7400 does support IPP (Inter-PHY-Protocol) which, at first glance, seems to enable them to co-exist with HD-PLC silicon). Note that the existence of ISP can be proved when it interoperates with a G.hn device, and G.hn devices are currently not in the market. Unless an interoperability / coexistence fest is organized with both IEEE 1901 silicon and G.hn silicon, this can't be completely proved or disproved.

While we dealt with coexistence in the previous section, interoperability is a more interesting. This is where Sigma Designs scores big. In addition to being one of the first vendors to demonstrate working G.hn silicon, they also add the icing on the cake by indicating interoperability with both HomePlug and HomePNA devices. This means that any G.hn device made with Sigma Designs's CG 5110 (demonstrated at CES 2011) will be able to exchange data with HomePlug AV and HomePNA devices.

Sigma Designs is not putting all its eggs in one basket either. Earlier this month, they introduced their second generation HomePlug AV silicon (CG 2210) compliant with the IEEE P1901 specifications. Despite having only a 200 Mbps PHY like the CG 2110, it includes a number of architectural improvements and reduced power consumption (being fabricated in the 55nm process, compared to the 90nm process used in the CG2110).

Final Words

G.hn is gaining momentum with silicon from multiple vendors. Marvell is the latest entrant to the game and it appears likely that they will win over all the UPA adopters with their G.hn offering. In the big picture, however, G.hn is up against HomePlug's huge install base. Irrespective of which standard wins out in the end, coexistence is an absoute necessity and interoperability would be nice to have. In this respect, Sigma Designs has covered all the bases with silicon for both HomePlug AV and G.hn. With today's silicon introduction, Marvell is betting big on G.hn winning. Will Marvell need to change its stance? Only time will tell.